Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
pg. 0 pg. 0
Addressing Data Accuracy and Information Integrity in mHealth using ML
by
Zaid Zekiria Sako Bachelor of Business (Business Information Systems)
Submitted in fulfilment of the requirements for the degree of
Master of Applied Science
Deakin University
August 2018
pg. 3
Abstract
Healthcare is a large, extensive, and complex system that aims to help restore
peoples’ health and assist patients where possible. A number of health services
make this possible through the delivery of care by medical professionals.
However, the constant pressures facing the healthcare systems around the
globe are making the delivery of healthcare services a challenge. The growing
number of aging population, the rising number of chronic diseases, the number
of cases of medical harm leading to quality issues in healthcare, and the
constant rise in healthcare costs, are making the delivery of care a challenge.
To address this challenge, technologies such as Mobile Health (mHealth) are
grabbing the attention of academics and medical professionals, to try and
establish methods of delivering healthcare services using this technology.
However, technologies do have some challenges that can become a barrier in
making mHealth a possibility. One area this research focuses on, is the Data
Quality of Information Systems, specifically in the context of mHealth, and the
Integrity of the information produced as a result of this data. This study involves
the use of a qualitative research method that examines diabetes data, a case
that represents one of the solutions mHealth is being deployed for. The
qualitative study searches for themes in the data to understand ways inaccurate
data can be captured and how Machine Learning techniques, can be deployed
to prevent erroneous data from being used in mHealth.
pg. 4
Table of Contents
Abstract ....................................................................................................................... 3
Table of Figures ........................................................................................................... 6
Table of Tables ............................................................................................................ 7
List of Abbreviations.................................................................................................... 8
Chapter 1 : Introduction ......................................................................................... 9
Chapter Summary ................................................................................................. 17
Chapter 2 : Literature Review ............................................................................. 18
2.1 Healthcare Delivery .................................................................................... 18
2.2 Challenges in Healthcare Delivery ........................................................... 26
2.3 IoT ................................................................................................................. 32
2.4 mHealth ........................................................................................................ 39
2.5 Rise of mHealth .......................................................................................... 45
2.5.1 Aging Population: ................................................................................ 45
2.5.2 Accessibility: ......................................................................................... 47
2.5.3 Self-management: ............................................................................... 47
2.5.4 Intervention: .......................................................................................... 48
2.5.5 Adherence: ........................................................................................... 49
2.5.6 Health promotion: ................................................................................ 50
2.5.7 Seeking health information online: ................................................... 50
2.5.8 mHealth Applications (Apps): ............................................................ 50
2.6 Chronic diseases, challenges and success factors .............................. 51
2.7 mHealth for chronic diseases ................................................................... 57
2.8 mHealth for Diabetes ................................................................................. 61
2.9 Patient – Clinician Interaction ................................................................... 67
2.10 Big data for Health .................................................................................. 73
2.11 Data Accuracy ......................................................................................... 80
2.12 Information Integrity ................................................................................ 86
2.13 Machine Learning ................................................................................... 89
Chapter Summary: ................................................................................................ 95
Chapter 3 : Research Methodology ................................................................... 97
3.1 Single Case Study ...................................................................................... 97
3.2 Data Collection ............................................................................................ 99
3.3 Data Sampling ........................................................................................... 101
3.4 Data Triangulation .................................................................................... 102
3.5 Data Analysis ............................................................................................ 102
3.5.1 Thematic Analysis ................................................................................. 103
pg. 5
3.5.2 Hermeneutic Analysis ........................................................................... 103
Chapter Summary: .............................................................................................. 105
Chapter 4 : Data Analysis and Findings ......................................................... 107
4.1 Domain Description and understanding of Diabetes .......................... 107
4.2 Describing the secondary data ............................................................... 109
4.3 Review of secondary data of patients with diabetes ........................... 111
4.4 The transformation of data before Machine Learning ......................... 114
4.5 Labelling the data ..................................................................................... 116
4.6 Defining the problem and understanding the data .............................. 117
4.7 Building the Model and Selection of Weka Machine Learning Tool . 118
4.8 Data Pre-processing ................................................................................ 120
4.9 Findings ...................................................................................................... 122
4.9.1 Thematic Analysis ............................................................................. 122
4.9.2 Machine Learning Outcomes .......................................................... 125
Chapter 5 : Discussion....................................................................................... 130
5.1 The different levels of Data Accuracy.................................................... 130
5.2 Understanding the human behaviour .................................................... 131
5.3 The role of metadata ................................................................................ 132
5.4 Contextual data and the inclusion of other metadata ......................... 133
5.5 Machine Learning and Artificial Intelligence ......................................... 134
5.6 Privacy of users and data security ......................................................... 136
Chapter Summary: .............................................................................................. 137
Chapter 6 : Conclusion ...................................................................................... 137
6.1 Answering the research question ........................................................... 138
6.2 Contribution to Theory and Practice ...................................................... 140
6.2.1 Contribution to Theory ...................................................................... 140
6.2.2 Contribution to Practice .................................................................... 141
6.3 Limitations .................................................................................................. 141
6.4 Future research directions ...................................................................... 142
6.5 Lessons Learnt.......................................................................................... 142
Chapter Summary: .............................................................................................. 143
References ............................................................................................................... 143
Appendix A ............................................................................................................... 164
Appendix B ............................................................................................................... 167
pg. 6
Table of Figures Figure 1.1
UN Statistics on Aging Population. Adapted from (UN, World Population Ageing, 2015)
p.10
Figure 1.2
Causes of premature deaths due to chronic diseases. Adapted from (WHO, 2016)
p.11
Figure 1.3
Statistics on mobile subscriptions trend around the globe. Adapted from (ITU, 2016)
p.12
Figure 1.4
Battling chronic disease using any form of mHealth technology. Outer circle adapted from (WHO, 2016) and inner circle shows ways of battling those diseases.
p.15
Figure 2.4.1
Source PwC Report, based on data from Economist Intelligence Unit 2012
p.43
Figure 2.6.1
Number of deaths by risk factors (Source: Our ln World In Data, 2018).
p.53
Figure 2.12.1
The Omaha System 2005 version (Adapted from The Omaha System Chart, 2018)
p.88
Figure 3.1.1
Research Design p.99
Figure 3.5.1
Analysis Design p.106
Figure 4.2.1
Example of the Secondary Data. Example is from Dataset 1 p.110
Figure 4.3.1
Structure of the raw data from Data20 p.113
Figure 4.3.2
Example of old data structure from Data1 before it was transformed into new format
p.114
Figure 4.3.3
The new structure of the data for Data1 p.114
Figure 4.5.1
Labelling of the data p.117
Figure 4.6.1
Data Samples from Diabetic Dataset p.117
Figure 4.7.1
Machine Learning Model p.119
Figure 4.7.2
Process of building a Machine Learning Model 1.2 p.119
Figure 4.7.3
Process of building the model 1.2 p.120
Figure 5.5.1
Application of Machine Learning in mHealth solutions p.134
pg. 7
Table of Tables Table 1.1
Classification of devices and sensors. Adapted from (Ade, 2012) p.13
Table 2.1.1
Delivery of Healthcare Services. Adapted from (Griffin et al., 2016)
p.19
Table 2.1.2
Comparison of the differences between traditional and PC hospital. Adapted from (Verzillo et al., 2018)
p.24
Table 2.3.1
Applications for Healthcare in the IoT. Adapted from (Laplante & Laplante, 2016)
p.33
Table 2.3.2
IoT Applications. Adapted from (Farahani et al., 2018) p.36
Table 2.4.1
Table 2.4.1 Applications of mHealth. Adapted from (Mirza, Norris, & Stockdale, 2008)
p.41
Table 2.6.1
Definition of Risk Factors according to AIHW, 2018 p.52
Table 2.6.2
Table 2.6.2 Some barriers, current initiatives and possible enhancements to general practice care for people with chronic disease. Adapted from (Harris & Zwar, 2007)
p.54
Table 2.7.1
SMS4BG Modules. Adapted from (Dobson et al., 2015) p.60
Table 2.8.1
Details of mobile application intervention for chronic disease management. Adapted from (Youfa et al., 2017)
p.63
Table 2.8.2
Examples of tailored text-messaging for self-management support for Blood Glucose (SMS4BG) (Dobson et al., 2015)
p.66
Table 2.9.1
Patient-clinician communication in hospitals (Source: safetiyandquality.gov.au, 2018)
p.69
Table 2.10.1
Business Intelligence vs Big Data, Adapted from (Trifu & Ivan, 2014)
p.74
Table 2.10.2
Lessons Learned in Applying Data to Drive Care. Adapted from (Jones, Pulk, Gionfriddo, Evans, & Parry, 2018)
p.79
Table 2.11.1
Definitions of Data Quality Elements. Adapted from ((WHO, 2014) and (Batini & Scannapieco (2016))
p.82
Table 2.11.2
Data Quality Dimension Framework. Adapted from (Talburt, 2010)
p.83
Table 2.11.3
HIT errors can lead to adverse events and patient harm. Adapted from (Magrabi et al., 2016)
p.84
Table 2.11.4
Source of Inaccurate Data. Adapted from (Olson, 2003) p.84
Table 2.12.1
Components of Omaha System p.89
Table 2.13.1
Role of Machine Learning in mHealth p.91
Table 2.13.2
Machine Learning for Healthcare: On the Verge of a Major Shift in Healthcare Epidemiology. Adapted from (Wiens & Shenoy, 2018)
p.94
Table 4.2.1
Codes and their explanations p.110
Table 4.4.1
List of A priori codes generated from the literature review p.115
6.1.1. Addressing the Element of Data Quality in this study. p.139
pg. 8
List of Abbreviations
AI Artificial Intelligence
AIHW Australian Institute for Health and Welfare
BI Business Intelligence
EHR Electronic Health Record
HIT Health Information Technology
IoT Internet of Things
ITU International Telecommunication Union
ML Machine Learning
mHealth Mobile Health
NCD Noncommunicable diseases
PC Patient-Centered
UN United Nations
WHO World Health Organization
pg. 9
Chapter 1 : Introduction Health, according to the World Health Organization (WHO) is ‘a state of which
complete physical, mental and social well-being and not merely the absence of
disease or infirmity’ (WHO, 2018). Healthcare systems focus on meeting the
health needs of a particular population (Glowik & Smyczek, 2015). These
essential systems are continuously exposed to pressures which lead to making
the delivery of healthcare services challenging.
Several factors inside and outside the healthcare systems, are creating those
challenges. Rising number of aging population, the ever-increasing costs of
health services, injuries due to quality and safety issues within the delivery of
healthcare, and chronic diseases, are making the delivery of healthcare a
challenge (Armstrong, Gillespie, Leeder, Rubin, & Russell, 2007). The aging
population is at an ever-increasing rate as people live longer and maintain a
healthier routine (Foscolou et al., 2018).
An instance of this increasing rate in aging population is felt in Australia as the
life expectancy at birth for people born in 2014 is 84.4 years for girls and 80.3
for boys, compared to 1890’s where life expectancy was 50.8 years for girls,
and 47.2 for boys (Australian Institute of Health and Welfare (AIHW), 2018).
The aging population in Australia, people that are aged 65 and over, is 3.7
million, that makes 15% of the total Australian population which stands at 24.4
million (AIHW, 2018). At a global level, statistics from the WHO indicate that the
aging population is continuing to rise as people are living longer with an
expectation of living beyond the age of 60 (WHO, 2018). This aging population
of people aged 65 or older will account for 1.5 billion worldwide by 2050 (WHO,
2018). The United Nations (UN) also demonstrate this rate of increase in Figure
1.1, where aging population around the world is increasing and the forecast for
the aging population continues to rise in proportion to population growth (UN,
2018). The effect of such increase is felt at the social, political and economic
levels (Peter, Stefan, & Markus, 1999).
pg. 10
Figure 1.1: UN Statistics on Aging Population. Adapted from (UN, World Population Ageing, 2015)
With aging comes the problem of people developing diseases as they get older
(Niccoli & Partridge, 2012). Common health problems that are developed with
aging are hearing loss, cataracts and refractive errors, back and neck pain and
osteoarthritis, chronic obstructive diseases, diabetes, depression and dementia
(WHO, 2018).
As life expectancy increases, the likelihood of people developing chronic
diseases increases as well. Chronic diseases, also referred to as
noncommunicable (NCD) diseases, is ‘a disease that cannot be cured and
never disappears’ as defined by Healey and Evans (2015, p. 299). Chronic
diseases according to WHO (2018), are the top four diseases that range from
cardiovascular diseases, cancers, respiratory diseases and diabetes. Reports
from the WHO indicate that chronic diseases are the leading cause of mortality
worldwide, where the number of deaths from 2012 are projected to increase
from 38 to 52 million by 2030 (WHO, 2018). People suffering from a chronic
disease might develop a certain limitation (Boccolini, Duarte, Marcelino, &
Boccolini, 2017). At the social level, people develop a limited capacity of
performing certain tasks and live life within a certain routine (Boccolini et al.,
pg. 11
2017). In Australia, at the economic level, people would become limited to
certain working conditions, and in 2002, there were more than 5 people of
working age to support every person aged over 65 (Australian Government,
2018). By 2042, there will be only 2.5 people of working age supporting each
person aged over 65 (Australian Government, 2018).
Despite people living longer, the chances of developing a chronic disease leads
to premature deaths and can reduce the quality of life while suffering from a
chronic disease (Malvey & Slovensky, 2014). Statistics from AIHW between
2011 and 2012 have shown similar figures presented by the WHO as about 5%
the of Australian adult population (accounting for approx. 917,000 people) have
self-reported diabetes, including sources from biomedical data (AIHW, 2018).
In the period between 2013 and 2014, diabetes was a factor in 9% of all
hospitalization (approx. 900,000 people) in Australia (AIHW, 2018). Figure 1.2
illustrates the causes of premature deaths and the contributing percentage of
each chronic disease to premature deaths.
Figure 1.2: Causes of premature deaths due to chronic diseases. Adapted from (WHO, 2018)
AIHW have reported that four major disease groups (chronic – cardiovascular,
oral health, mental disorders, and musculoskeletal) have cost the Australian
pg. 12
healthcare system an estimated $27 billion in 2008-09 (AIHW, 2018). In the US,
the healthcare expenditure has risen from 5% in 1960 to 19% (2015 figure) and
is expected to reach 40% of GDP by 2040 (Hosseini, 2015).
All modern healthcare systems are facing common challenges caused by
globalization, medical and technological progress, and demographic change
(Smith, 2011). This requires healthcare systems to adjust and change their
ways of how provisioning of a service is organized (Smith, 2011). The presence
of mobile phone technology presents an opportunity for the healthcare
challenges to be met through innovative solutions. The number of mobile phone
subscriptions as per the 2015 statistics released by the International
Telecommunication Union (ITU) (See Figure 1.3), is estimated to be 7 billion
worldwide (ITU, 2018) and it is expected to reach 24 billion devices by 2020
(Serhani, Harous, Navaz, Menshawy, & Benharref, 2017). The growing number
of mobile phone subscriptions is driven by peoples’ needs and habits, and it is
not imposed (Taruna, 2014).
Figure 1.3: Statistics on mobile subscriptions trend around the globe. Adapted from (ITU, 2018)
pg. 13
This presents an opportunity for mobile phone devices to be used for the
management of health and as an intervention in the rising number of chronic
diseases, as half of the smartphone owners frequently browse for health
information online and monitor their health using mobile health applications
(Fox & Duggan, 2012). As such, the new role for mobile phones is mobile health
(mHealth). The definition of mHealth is the use of portable devices such as
smartphones and tablets to improve health (Hamel, Cortez, Cohen, &
Kesselheim, 2014). mHealth has enabled people to play an active role in
managing their health rather than being the passive object when seeking
treatment during traditional methods (Niilo, Ilkka, & Elina, 2006).
This technology does not only enable self-management but can also reduce
healthcare costs (LeRouge & Wickramasinghe, 2013). For instance, one of the
methods of reducing healthcare costs is through self-care (Wickramasinghe &
Gururajan, 2016). Home self-care for example, can reduce the healthcare costs
by allowing follow-up care by nurses via telephone calls and home visits that
significantly decrease hospital readmissions (Schatz & Berlin, 2011). The
advances in sensor technology such as heart rate, respiratory and blood
pressure sensors, along with mobile phone devices, have allowed patients to
self-monitor their health before conditions deteriorate and risk being readmitted
to hospital (Tarassenko & Clifton, 2011).
Table 1.1 Classification of devices and sensors. Adapted from (Ade, 2012).
Form Description Examples
Traditional Medical devices connected to a mobile
phone via Bluetooth and which use the
phone as a data gateway
Blood pressure monitors,
Precision weight scales,
Spirometers
Implantable Implantable medical devices embedded
just below the surface of the skin and
which use a small transmitter to send
data to a gateway attached to the
patient or located in their vicinity.
Implantable devices are typically used
to monitor disease conditions such as
pg. 14
chronic diabetes where 24 hour a day
constant monitoring is required
Wearable Body sensors and wearable devices
contain one or more small, embedded
sensors capable of monitoring different
physiological and physical parameters.
BodyMedia Armband and
monitoring system
Ingestible Small micro-transmitters in chemical
agents, which react with the contents of
the stomach to generate electrical
signals which are then transmitted via a
data hub to a medical station
Proteus ingestible
sensors and monitoring
system
Mobile
Peripherals
Devices that can be plugged directly
into a mobile phone and which turns the
mobile phone into a medical instrument
AgaMatrix device
manufactured by Sanofi
Aventis and which is used
to monitor diabetes
Mobile
Apps
Applications that are developed to
replicate functionality that is normally
provided by a dedicated medical device
iPhone Stethoscope app
that records a heart
rhythm and plays back a
cardiogram
Individuals can self-monitor the status of their health using any form of mHealth
technology as presented in Table 1.1 which allow for different uses and
applications. The variety of mHealth solutions does not only introduce
innovative solutions but can also cater for patients with different needs and
diseases (See Figure 1.4) that make battling chronic diseases much more
powerful by allowing patients to be proactive in managing their health. The
different forms of mhealth solutions also play an important role in extending
healthcare to be delivered at the right time by bypassing all the geographical
barriers that patients might have when trying to access healthcare services.
pg. 15
Figure 1.4: Battling chronic disease using any form of mHealth technology. Outer circle adapted from (WHO, 2016) and inner circle shows ways of battling those diseases
This technology is changing the dynamics of healthcare by allowing for
evidence to be collected at the population scale in different time scales, track
activities and outcomes in real time, and even extend the patient to be involved
in their own treatment (O'Riordan & Elton, 2016).
While smartphones have a new role to play in the effective management of
health and diseases, the technology must be clear of medical errors. Errors
have been defined as active and latent (Kalra, 2011). Active errors are those
that produce immediate events and involve operators of complex systems such
as errors by healthcare professionals in medicine (Kalra, 2011). Latent errors
are those that result from factors that are inherent in the system, such as
excessive workload, insufficient training, and inadequate maintenance of
equipment (Kalra, 2011).
pg. 16
Errors in the medical field belong to a number of domains such as development
and use of technologies, ergonomics, administration, management, politics and
economics (Vincent, 2010). A common root cause is human error, where errors
are of omission (forgetting to do something) and commission (intentionally
doing something that is not meant to be done) (Health informatics: improving
patient care, 2012). The famous report from the Institute of Medicine (IOM) titled
“To Err Is Human” that was published in 1999, estimated that approximately
98,000 people die each year in hospital due to preventable medical errors
(Kohn, Corrigan, & Donaldson, 2000). In 2016, the Johns Hopkins (2016) report
of medical errors showed that the number of medical errors has doubled, as it
estimated that 250,000 people die each year due to medical errors.
Examples of medical errors from recent literature include when patients both
with and without a chronic disease have suffered a medication error (Nuovo,
2010), errors in discharge prescriptions due to incomplete and inadequate
prescriptions (Murray, Belanger, Devine, Lane, & Condren, 2017), and data
entry errors resulting in hospital staff being incorrectly tested for Hepatitis B, yet
when the correct test was carried, it was found the staff lacked immunity
(Vecellio, Maley, Toouli, Georgiou, & Westbrook, 2015). These examples
demonstrate few instances of how and when medical errors occur, and the
impact medical errors have on both healthcare and people.
With the introduction of technology in the healthcare domain, a new and
emerging contributor to medical errors is technology, whilst human errors are
still occurring (Jenicek, 2010). Jenicek (2010) defines technology errors in
medicine as errors that relate to data and information recording, processing,
and retrieval caused by information technology and its uses (information
technology inadequacy and failure). Modern healthcare patients are becoming
more of consumers than a patient due to the great control over the decisions
that affect their healthcare (Konschak & Jarrell, 2010). Safe clinical care
requires quality data and documentation (Medhanyie et al., 2017). Patient
safety as defined by Zacher (2014, p.7), and one that’s based on World Alliance
for Patient Safety is “the reduction of risk of unnecessary harm associated with
healthcare to an acceptable minimum”.
pg. 17
The use of mobile phone technology for managing health has its own set of
challenges and complexities, such as accuracy, integrity, privacy, security and
confidentiality. The technology can be a great tool that can benefit healthcare,
people and society. To keep the momentum of continuous mhealth
developments and innovation in healthcare, this research addresses the key
areas of mHealth that are the foundational features of the technology.
First, is an overview of Healthcare delivery along with the challenges in
healthcare delivery. Second is a review of Internet of Things (Iot) followed by
mHealth and Rise of mHealth. Third, is a review of the challenges in the delivery
of healthcare. Third is how mHealth is used for chronic diseases, with focus on
mHealth for diabetes. Fourth, is a review of human interaction with technology
and how errors might occur during the interaction stage. Fifth is a look at Big
Data and how Big Data occurs in the health domain. Sixth is data quality and
the accuracy dimension of data is reviewed with definitions underlying the key
components of accurate data. Seventh, a review of Information Integrity, and
what constitutes Information Integrity after data is processed, along with the use
of Omaha Care Plan to assist in understanding the generation of information.
Last, is a review of Machine Learning and how data is solely the underlying
source of insight and knowledge that is extracted using Algorithms, with a view
of Machine Learning in healthcare.
Chapter Summary
Constant pressures in the healthcare domain are making the delivery of
healthcare services more challenging. Some of these challenges are being
solved using technologies such as mHealth, that have the potential of lowering
healthcare costs while delivering more care to patients simultaneously
empowering patients to self-manage their health. While the opportunity is there,
the technology must be free of any errors that can potentially contribute back to
medical errors.
pg. 18
Chapter 2 : Literature Review The literature review includes a review of healthcare delivery, the challenges in
healthcare delivery, IoT, rise of mHealth, mHealth technology, chronic disease
challenges and success factors, mHealth for chronic diseases, mHealth for
Diabetes, Patient – Clinician Interaction, Big data for Health, Data Accuracy,
Information Integrity, and Machine Learning.
2.1 Healthcare Delivery
Healthcare is a system involving inputs (finance, workforce), processes, outputs
and outcomes (Willcox et al., 2015), that is delivered by trained an licensed
professionals in medicine, nursing, dentistry, pharmacy, and other allied health
providers (Griffin et al., 2016). The quality and accessibility of healthcare is
different from one country to another, and it is shaped by health policies that
are in place (Griffin et al., 2016). The outputs and outcomes of the healthcare
system include individual or person-level outputs (patients treated) and
outcomes (improved quality of life) and wider outputs/outcomes (research
outputs, strong communities, changed environments) (Willcox et al., 2015).
Health outputs and health outcomes may not be distributed evenly across all
members of society (Willcox et al., 2015). The healthcare system can be
evaluated in terms of its impact on equity, quality, efficiency and acceptability
(Willcox et al., 2015).
Healthcare systems are generally of two types, welfare-based, a public system
by the government that provides healthcare to all citizens, and a market-driven,
a private system that is run by private providers and is paid for by citizens
(Willis, Reynolds, & Keleher, 2016). The Australian healthcare system for
example, provides population health prevention inclusive of general practice
and community health, emergency health services and hospital care, and
rehabilitation and palliative care (healthdirect, 2018). The funding of such
system is through the government such as the Australian Medicare, or through
the private health system (Australian Government, 2018).
pg. 19
The 3 fundamental objectives of a Healthcare System according to the WHO,
are (Coady, Clements, Gupta, & International Monetary, 2012, p. 3):
• Improving the health of the population it serves
• Providing financial protection against the costs of ill-health; and
• Responding to people’s expectations.
The delivery of healthcare services is dependent on the conditions or need of
care. Table 2.1.1 highlights the different types of care, the delivery method, how
it is provided, and the condition that instigates the type of care. Table 2.1.1 also
highlights the complexity and the components of healthcare delivery.
Table 2.1.1: Delivery of Healthcare Services. Adapted from (Griffin et al., 2016).
Type Delivery Focus Providers Conditions/Needs
Primary
Care
• Day-to-Day
healthcare
• Often the first
point of
consultation for
patients
• Primary care
physician, general
practitioner, or
family or internal
medicine physician
• Pediatrician
• Dentist
• Physician assistant
• Nurse practitioner
• Physiotherapist
• Registered nurse
• Clinical officer
• Ayurvedic
• Routine
checkups
• Immunizations
• Preventive care
• Health
education
• Asthma
• Chronic
obstructive
pulmonary
disease
• Diabetes
• Arthritis
• Thyroid
dysfunction
• Hypertension
Vaccinations
• Oral health
• Basic maternal
and child care
pg. 20
Urgent
Care
• Treatment of
acute and
chronic illness
and injury
provided in a
dedicated walk-
in clinic
• For injuries or
illnesses
requiring
immediate or
urgent care but
not serious
enough to
warrant an ER
visit
• Typically do not
offer surgical
services
• Family medicine
physician
• Emergency
medicine physician
• Physician assistant
• Registered nurse
• Nurse practitioner
• Broken bones
• Back pain
• Heat exhaustion
• Insect bites and
stings
• Burns
• Sunburns
• Ear infection
• Physicals
Ambulatory
or
outpatient
care
• Consultation,
treatment, or
intervention on
an outpatient
basis (medical
office,
outpatient
surgery center,
or ambulance)
• Typically does
not require an
overnight stay
• Internal medicine
physician
• Endoscopy nurse
• Medical technician
• Paramedic
• Urinary tract
infection
• Colonscopy
• Carpal tunnel
syndrome
• Stabilize patient
for transport
Secondary
or acute
care
• Medical
specialties
typically needed
for advanced or
acute conditions
• Emergency
medicine physician
• Cardiologist
• Urologist
• Dermatologist
• Emergency
medical care
• Acute coronary
syndrome
• Cardiomyopathy
pg. 21
including
hospital
emergency
room visits
• Typically not the
first contact with
patients; usually
referred by
primary care
physicians
• Psychiatrist
• Clinical psychologist
• Gynecologist and
obstetrician
• Rehabilitative
therapist (physical,
occupational, and
speech)
• Bladder stones
• Prostate Cancer
• Women’s health
Tertiary
care
• Specialized
highly technical
healthcare
usually for
inpatients
• Usually patients
are referred to
this level of care
from primary or
secondary care
personnel
• Surgeon (cardiac,
orthopedic, brain,
plastic, transplant,
etc.)
• Anesthesiologist
• Neonatal nurse
practitioner
• Ventricular assist
device coordinator
• Cancer
management
• Cardiac surgery
• Orthopedic
surgery
• Neurosurgery
• Plastic surgery
• Transplant
surgery
• Premature birth
• Palliative care
• Severe burn
treatment
Quaternary
care
• Advanced levels
of medicine that
are highly
specialized and
not widely
accessed
• Experimental
medicine
• Typically
available only in
a limited
number of
• Neurologist
• Ophthalmologist
• Hematologist
• Immunologist
• Oncologist
• Virologist
• Multi-drug
resistant
tuberculosis
• Liver cirrhosis
• Psoriasis
• Lupus
• Myocarditis
• Gastric cancer
• Multiple
myeloma
• Ulcerative colitis
pg. 22
academic health
centers
Home and
community
care
• Professional
care in
residential and
community
settings
• End-of-life care
(hospice and
palliative)
• Medical director
(physician)
• Registered nurse
• Licensed practical
nurse
• Certified nursing
assistant
• Social worker
• Dietitian or
nutritionist
• Physical,
occupational, and
speech therapists
• Post-acute care
• Disease
management
teaching
• Long-term care
• Skilled nursing
facility/assisted
living
• Behavioural
and/or
substance use
disorders
• Rehabilitation
using
prosthesis,
orthotics, or
wheelchairs
Healthcare, as a system, is complicated, and the complexity in the system
arises from six interconnected levels— the patient, the population, the team,
the organization, the network, and the political and economic environments
(PPTONE) (Griffin et al., 2016). Aspalter, Pribadi, and Gauld (2017)
demonstrate this complexity in healthcare by illustrating the process of care
when an individual seeks health services through their general practitioner. The
example is that individual practitioners, such as primary care physicians
working in solo practice and their individual patients, are inevitably a part of a
broad system of care. The patient with a condition that the solo primary medical
practitioner is unable to diagnose, and treat will be referred on to a provider with
a higher level of training, qualification and specialization. That provider may
then require additional input from others, such as those with specific technology
required to diagnose and to deliver effective treatment. Of course, prescribing
is central to most patient consultations, and the drugs themselves are often
dispensed by independent pharmacists to ensure there is a separation between
pg. 23
prescribing and dispensing. Where such separation does not exist (as is the
case in some Asian countries), this provides a potential incentive for the
prescribing doctor to indicate medicines that generate a higher personal profit;
it can also mean that patients do not necessarily obtain the expert advice and
second opinion often provided by pharmacists. Many patients require
hospitalization, either through referral from a primary care physician or due to
an accident or emergency that requires treatment or ongoing observation. At
any of these points, information on the patient is critical including medical
history, current treatment regime and any test or other diagnostic results. For
this reason alone, information systems are a fundamental foundation for
healthcare systems (Aspalter, Pribadi, & Gauld, 2017).
Healthcare systems around the world operate differently and face different
challenges (Baldwin, 2011). The US healthcare system for instance, is
fragmented and currently designed to support billing rather than care
coordination and continuity (Gupta & Sharda, 2013). Part of this complexity in
the US healthcare system is due to low efficiencies, poor technology interfaces,
communication gaps, and high costs (Gupta & Sharda, 2013). Healthcare
systems also differ between developing and developed countries. In developing
countries, access to universal healthcare is deemed urgent while in developed
countries, governments focus on increasing access to additional private
healthcare coverage to supplement government-run universal coverage
(Newswire, 2018).
WHO (2018) explains that a “good health system delivers quality services to all
people, when and where they need them. The exact configuration of services
varies from country to country, but in all cases, it requires a robust financing
mechanism; a well-trained and adequately paid workforce; reliable information
on which to base decisions and policies; well-maintained facilities and logistics
to deliver quality medicines and technology”.
Ensuring life is the core goal of the health system as explained by Knickman,
Kovner, and Jonas (2015). WHO (2018) describes a well-functioning health
system as “A health system working in harmony is built on having trained and
pg. 24
motivated health workers, a well-maintained infrastructure, and a reliable
supply of medicines and technologies, backed by adequate funding, strong
health plans and evidence-based policies”. The healthcare systems comprise
of multiple components that allow people access to health services and care.
Healthcare systems are fragmented, disorganized and unaccountably variable
as explained by Yih (2014).
Healthcare systems are also experiencing a shift in the way care is delivered.
New models such as Patient-centered and Patient Direct healthcare, are an
example of a shift in healthcare system that moves from provider focused to
more patient first model (Konschak, Nayyar, Morris, & Levin, 2013). The
redesign of the care delivery towards patient centered model (PC) according to
Verzillo, Fioro, and Gorli (2018, p. 3) is an “attempt to redesign the care delivery
process by shaping the structures and processes involved in delivering hospital
care according to the needs of the patients. In the traditional hospital models,
patients are admitted under individual specialist clinicians, who keep them or
transfer them to the care of another clinician.”.
Table 2.1.2: Comparison of the differences between traditional and PC hospital. Adapted from (Verzillo et al., 2018).
Functional hospital
configuration
More recent
innovations:
converging patterns
towards PC hospitals
Organizational
model/care delivery
model
Functional/divisional
model
Lean organization
/process-oriented model
Organizational unit:
patients’ care needs and
the relationship among
specialties
Specialty-based units.
Practitioners (doctors and
nurses) are grouped into
semi-autonomous units
depending on their
specialty of belonging
Multi-specialty units. Units
are aggregated in
accordance with patients’
clinical and assistential
needs. Doctors might
treat patients located in
different units and nurses
pg. 25
might assist patients with
different pathologies
Model of care Functional nursing
(nurses’ task-oriented job:
each nurse is specialized
in a single care activity)
Modular nursing (nurses
are responsible for
the overall assistential
practices required
by small groups of
patients within the
ward)
Use of resources Separated resources
(beds, operating rooms,
equipment, nursing staff,
other staff) devoted to the
individual specialties
Resource pooling:
resources are shared by
all the functional
specialties regrouped
Managerial roles Head physicians in
charge of their
departments
Bed manager/case
manager (as
distinguished by the
clinical activity) for
centralized operation
management
Physical environment Hospitals are built around
fixed and focused spaces,
with often isolated wings
Newly built hospitals are
designed to maximize
resource pooling and
patient grouping, flexibility
and modularity of spaces
Table 2.1.2 is a comparison of the differences between traditional and PC
Hospital. Situations where providers talk to patients, than with them; blame
patients when treatments fail; have little expectation or trust that patient will
participate in their own care; and exhibit little patience with patients who
question the provider’s approach, research their own conditions, or ask for more
time or information (Konschak et al., 2013).
The strategic imperative behind PC is based on patient centeredness with a
focus on individual responsibility and the promotion of prevention and self-
pg. 26
management of diseases (Moretta Tartaglione, Cavacece, Russo, & Cassia,
2018). PC models engage patients in care-related decisions, being involved in
and adhering to their plan of care, that enhances the level of satisfaction with
care and improves outcomes (Sidani et al., 2018). The new PC models have
been trialed and implemented in some instances. A study that focused on low
back pain, conducted a pilot randomized controlled trial that combined primary
care and chiropractic care to produce a patient-centered model, resulted in
better perceived improvement and patient satisfaction (Goertz et al., 2017).
Good health is closely linked to economic growth through higher labor
productivity, demographic growth, and higher educational attainment
(Ngwainmbi, 2014).
While healthcare systems aim to maintain or restore the human body through
treatments and other means of disease prevention (Griffin et al., 2016), yet
there are a number of challenges that make delivery of healthcare difficult. The
Australian healthcare system, like many other healthcare systems around the
world, is under pressure as the demand for more health services increases as
a result of aging population, rising number of chronic diseases and high
consumer expectation (Carlisle, Fleming, & Berrigan, 2016). These issues
create challenges in delivering healthcare services to people and for population
at large.
2.2 Challenges in Healthcare Delivery
Healthcare systems are facing many challenges as a result of aging
populations, chronic diseases, quality of care, patient safety and rising
healthcare costs (Armstrong et al., 2007).
Population aging is defined by (Rowland, 2012) as “an increase in the
proportion of the population in older ages”. The aging population is currently
progressing faster than previously predicted as the number of older population
is bypassing the number of birth rates (Rowland, 2012), which results in
changes in demography (Komp & Johansson, 2015). In Western Europe,
fertility rate is low hence the aging population is increasing in Europe, compared
pg. 27
to developing countries (Bengtsson, 2010). The increase in aging population
is leading to elderly people requiring more care compared to the younger
generation. An example of this is In the United Kingdom, where many of those
admitted to the hospital in 2016 were aged 65 and over (Gabriel, 2017). The
length of stay for patients between the age of 65-74 was 6.5 days, 8.3 for people
between 75-84, and 10.1 days for individuals over the age of 85 (Gabriel, 2017).
These numbers indicate the duration of care required for older people, which
leads to adding pressure on healthcare systems to deliver more care and strain
the funding of healthcare. Australia’s healthcare expenditure is expected to rise
as a result of the aging population requiring more services and care (Duckett &
Willcox, 2015).
Aging does not only pose a financial threat to healthcare systems but is also
changing and reforming policies. Superannuation in Australia has been
redesigned over the last 20 years, with ongoing debate about retirement
incomes policy, the establishment of a Commissioner on Age Discrimination
and the ‘rights’ approach, and consumer-led directions in the Living Longer
Living Better reforms of aged care, as Kendig, McDonald, and Piggott (2016)
explain the reform.
With aging comes the problem of people developing diseases as they get older
(Kennedy et al., 2014), as chronic diseases are common amongst older
persons where 65% of U.S. men and 72% of U.S. women above the age of 65
had two or more chronic illnesses in 2010 (Gaugler, 2015). Caring for patients
with chronic diseases and delivering care is an ongoing challenge (O'Loughlin,
Mills, McDermott, & Harriss, 2017) as it results in expensive bills, long term-
care and may involve multiple medical specialties (Klein & Neumann, 2008). In
England, 15.4 million people live with a chronic condition, which consumes 50%
of all the general practitioner appointments, 65% of all outpatient appointments,
and over 70% of all inpatient bed days (Saxton, 2011). The average number of
times a person sees a primary care provider is 2 and number of specialists is 5
per year, compared to a patient with chronic diseases who is likely to see 16
health professionals (Grossmann et al., 2011).
pg. 28
The human rights and health section from WHO (2018) state that Quality is a
key component of Universal Health Coverage and includes the experience as
well as the prescription of healthcare. Quality health services according to WHO
(2018) should be:
• Safe – avoiding injuries to people for whom the care is intended
• Effective – providing evidence-based healthcare services to those who need
them;
• People-centred – providing care that responds to individual preferences,
needs and values;
• Timely – reducing waiting times and sometimes harmful delays.
• Equitable – providing care that does not vary in quality on account of gender,
ethnicity, geographic location, and socio-economic status;
• Integrated – providing care that makes available the full range of health
services throughout the life course;
• Efficient – maximizing the benefit of available resources and avoiding waste
However, medicine is a field that is practiced by humans and the risk of humans
making an error is likely to occur (Kelsey, 2016). Patient safety according to
Jara, Zamora, and Skarmeta (2014) “is one of the most important issues in
worldwide healthcare”. Patient safety according to the WHO, is the 14th leading
cause of global disease burden (WHO, 2018). It is estimated that out of the 421
million annual hospitalisations worldwide, around 42.7 million bad outcomes are
a result of poor care (WHO, 2018). Noncompliance is an example of how
patients’ safety can be compromised and it occurs when a patient neglects to
take the prescribed dosage at the recommended times or decides to stop the
treatment without consulting their physician (Jara et al., 2014). A primary
responsibility of healthcare providers, facilities and systems is to do no harm
and do everything to ensure that the benefits of an intervention outweigh its
risks and deleterious effects as explained by Organisation for Economic Co-
operation and Development (2018). Drug compliance, Adverse Drug Reaction,
and the elimination of medical errors are important for improving patient care
pg. 29
(Jara et al., 2014). Harm in healthcare comes at a cost and it is one of the
reasons for the increase in healthcare costs (Van Den Bos et al., 2011).
The cost of healthcare delivery is rising and there are several factors
contributing to this increase (Dyas, Lovell et al. 2018). The relationship between
economic growth and health, have an effect on the economic and human
growth, such that malaria and Human immunodeficiency virus/ acquired
immune deficiency syndrome (HIV/AIDS) and other infectious diseases cause
a slow growth in the economy and human development (Ngwainmbi, 2014).
Patient safety for instance, costs around 42 billion annually, and this figure is
without the inclusion of lost wages, productivity, or healthcare costs
(WHO,2018). In 2014, the National Health Service (NHS) in England paid more
than 1 billion in legal claims (Kelsey, 2016). The aging population is another
variable that contributes to the increasing costs of healthcare as people grow
older, they require more care that results in increased costs (Howdon & Rice,
2018). Chronic diseases also account for a large portion of healthcare cost as
they result in significant economic burden because of the combined effects of
health-care costs and lost productivity from illness and death (AIHW, 2018).
Estimates based on allocated health-care expenditure indicate that the 4 most
expensive disease groups are chronic—cardiovascular diseases, oral health,
mental disorders, and musculoskeletal—incurring direct health-care costs of
$27 billion in 2008–09, which equates to 36% of all allocated health expenditure
(AIHW). Patient hospitalisation, also contributes to the increasing costs of
healthcare expenditure (Stey et al., 2015).
With the aforementioned challenges of healthcare, issues such as accessibility
and efficiency also create challenges in healthcare. According to Barjis,
Kolfschoten, and Maritz (2013, p. 1) “efficiency is becoming a buzzword in
discussion with both healthcare managers and decision makers. Behind the
term there is enormous complexity; from healthcare processes management to
decision making, policy, supporting technology, innovation, and socio-cultural
and economic realities — all are interwoven and interrelated into the equation
of efficiency”.
pg. 30
Access to healthcare is another challenge in healthcare systems for individuals
wanting to obtain health services. In Australia, many people find it difficult to
access health services for a variety of reasons (healthdirect, 2018). Groups that
are affected by such access barrier are those with specific needs, such as those
with complex and chronic health conditions who use health services frequently,
those in regional and rural areas, and Aboriginal and Torres Strait Islander
peoples (healthdirect, 2018).
One factor is the availability of health services and health professionals. There
are clear differences in access to services depending on where you live.
According to AIHW, there are lower rates of doctor consultation and generally
higher rates of hospital admission in regional and remote areas compared to
major cities. 70% of callers to the Australian after-hours GP helpline service do
not have access to a doctor after hours (evenings, weekends and public
holidays) (healthdirect, 2018). The barriers to access healthcare (healthdirect,
2018) are:
• Geographic locations: Health call centres are an effective way for people to
access healthcare information and advice, without time or geographic
restrictions. Otherwise, they would have little option but to visit their nearest
hospital, which could be hours away.
• Language: To achieve the best health outcomes, people need to find a
health provider they can communicate with and trust. There may be a lack
of services and information available in languages other than English, or a
lack of culturally appropriate services and information.
Another example of accessibility issue in healthcare is oral healthcare. Access
to oral healthcare is essential to promoting and maintaining overall health and
well-being (Committee on Oral Health Access to et al., 2011). When individuals
are able to access oral healthcare, they are more likely to receive basic
preventive services and education on personal behaviours (Committee on Oral
Health Access to et al., 2011). They are also more likely to have oral diseases
detected in the earlier stages and obtain restorative care as needed (Committee
pg. 31
on Oral Health Access to et al., 2011). In contrast, lack of access to oral
healthcare can result in delayed diagnosis, untreated oral diseases and
conditions, compromised health status, and, occasionally, even death.
Unfortunately, access to oral healthcare eludes many Americans.
A significant proportion of the U.S. population is not adequately served by the
current oral healthcare system, and millions of Americans have unmet oral
healthcare. This is especially true for the nation’s vulnerable and underserved
populations. Commonly studied populations include but are not limited to
(Committee on Oral Health Access to et al., 2011) :
• Racial and ethnic minorities, including immigrants and non– English
speakers
• Children, especially those who are very young
• Pregnant women
• People with special healthcare needs
• Older adults; Individuals living in rural and urban underserved areas
• Uninsured and publicly insured individuals
• Homeless individuals
• Populations of lower socioeconomic status
For example, in 2009, 4.6 million children did not obtain needed dental care
because their families stated that they could not afford it and people with
disabilities are less likely to have seen a dentist in the past year than people
without disabilities (Committee on Oral Health Access to et al., 2011).
Limited access to healthcare impacts quality of life and can lead to poorer health
outcomes (healthdirect, 2018). Charness, Demiris, and Krupinski (2011)
explain that the increase in the aging population, coupled with chronic diseases,
mean that there will be a great need for cost effective healthcare provisioning
in many countries. A way of overcoming these challenges is via the use of
technology that enables accessibility to healthcare, improves quality of care,
can reduce costs and ensure patient safety. Well-designed technology and
good decision making models can improve safety, efficiency and reduced costs
pg. 32
(Gupta & Sharda, 2013). The Internet of Things (IoT) is a technology that is
enabling communications between devices, connect people with devices and
services that can enhance and simplify daily lives and work.
2.3 IoT
Traditionally, modern Information and communication technologies (ICT) were
adopted in healthcare systems for the purpose of promising solutions for
efficiently delivering all kinds of medical healthcare services to patients, known
as e-health, such as Electronic Health Records (EHR), telemedicine systems,
and personalized devices for diagnosis (Qi et al., 2017). The use of consumer
health informatics technology (CHIT) for self-managing chronic diseases is
expected to help healthcare consumers assume greater responsibility for
managing their health; support the exchange of information among patients,
caregivers, and healthcare providers; and facilitate the patient–provider
partnership (LeRouge & Wickramasinghe, 2013).
Today’s technology is far more advanced than what it was a century ago.
Supercomputers, tiny sensors, mobile phones, wearable and implantable, and
Internet of Things (IoT), are all providing solutions to numerous challenges in
different industries, including healthcare. A technology that is enabling
connectivity amongst devices is IoT. IoT as defined by Hussain (2017, p. 1) is
“interconnection of things”, that is used to sense and report real world
information. IoT allows for people and objects in the physical world as well as
data and virtual environments to interact with each other (Verma & Sood, 2018).
The development of IoT was proposed by Ashton and Brock of Massachusetts
Institute of Technology (MIT) and later the term IoT was coined in 2005 by the
ITU (Y. Yin, Zeng, Chen, & Fan, 2016). IoT has since seen a strong trend in
adoption and solutions being developed using the technology. This great
intersection of small devices and healthcare presents an opportunity for the
technology to be introduced in healthcare and be used to break some of the
challenges in healthcare. IoT is a field that combines systems that may include
but not limited to embedded systems, electronics, wireless sensor networks,
pg. 33
communication networks, and computing paradigms (Baloch, Shaikh, & Unar,
2018). IoT serves as an umbrella above these technologies, which diversifies
its applications area. IoT applications include smart cities, smart homes/smart
buildings, environment monitoring, smart business and product management,
emergency response systems, intelligent transportation, security and
surveillance, energy and industrial automation, and healthcare (Baloch et al.,
2018).
IoT consists of four elements (Patil, 2017):
1. Things: Any physical thing, such as line-of-business assets, including
industry devices or sensors.
2. Connectivity: Those things that have connectivity to the internet.
3. Data: Those things that can collect and communicate information – this
information may include data collected from the environment.
4. Analytics: The analytics that come with the data produce insight and enable
people or machines to take actions that drive business outcomes.
The advances in IoT technology have resulted in the development of several
IoT applications in different domains, including the healthcare domain. These
advances in IoT technology are leading to the design of smart systems that
support and improve healthcare and biomedical-related processes (Catarinucci
et al., 2015). Automatic identification and tracking of people and biomedical
devices in hospitals, correct drug–patient associations, real-time monitoring of
patients’ physiological parameters for early detection of clinical deterioration are
a few examples of the possible solutions that can be created using IoT
(Catarinucci et al., 2015). Table 2.3.1 illustrates different examples of IoT in
healthcare, the application of IoT, how the sensors work, and how information
is collected as described by (Laplante & Laplante, 2016).
Table 2.3.1: Applications for Healthcare in the IoT. Adapted from (Laplante & Laplante, 2016).
Application Solutions using IoT
Bulimia
(Eating
Disorder)
Sensors can be used for people suffering from bulimia at both
hospital and home. In a patient’s room (hospital settings), sensors
can be used to detect increased body temperature or blood pressure,
or even the odour of vomit. Outside of hospital, the sensors can be
pg. 34
used to detect exercise abuse such as excessive cardio training or
accelerated walking activity as compared to walking at a normal pace.
All the information collected from these sensors can provide valuable
information in the diagnosis and management of the illness.
Alzheimer’s
Disease
IoT sensors can be deployed for people suffering from Alzheimer’s
disease. The IoT sensors can be used to track people’s geolocation
to prevent them from wandering or prohibit unwanted mobility
behaviours. Often, patients with Alzheimer’s suffer from comorbidities
with other diseases, such as hypertension (high blood pressure),
macular degeneration, or diabetes. Therefore, appropriate
interconnected devices could capture data for monitoring the unique
signs and symptoms of these conditions.
Safety and
Violence
Safety and violence are real issues in healthcare today. There are
numerous accounts of horizontal violence—for example, nurse
against nurse—but also of violence from visitors or family toward
healthcare providers or patients. Despite healthcare institutions
equipped with video surveillance systems, IoT could be deployed as
another measure of detecting violence and implement a zero-
tolerance policy. For example, tracking the movement of staff,
patients, and visitors could provide warnings of aberrant or
threatening behaviour. Biometric sensors could be used to detect
signs of aggression or stress in people who are entering or reside in
these settings.
Monitoring
inside the
hospital
IoT can also be used to track equipments inside the hospital. IoT can
be used by staff to try to keep certain equipment, such as an IV pump
or oxygen tanks, in their unit for future use. In a hospital, scarce
shared equipment such as EKG machines, IV pumps, and patient-
controlled analgesia (PCA) medication pumps could be tracked via
an IoT. In addition, the use of such equipment would be of interest to
individual units and administration—as well as insurance companies,
including Medicare—in documenting the need for additional
equipment. An IoT could also be used to monitor equipment that
needs to be refilled or calibrated, such as oxygen tanks, and to alert
staff of such situations.
Acute/Long
term care
IoT can be used for tracking supplies that can be scanned using low-
cost RFID or bar code tags for acute or long-term care setting, making
pg. 35
it easy to make charges to a patient’s account. Such supplies can
also be tracked using an IoT as they are either checked out from a
repository or administered to a patient. In some cases, where an
RFID tag is used, an item could be located more quickly, for example.
Likely trackable items include one-time use supplies, such as
dressings, catheters of different types, and personal care items. In a
home setting, medical supplies could be marked with RFID tags to
monitor use and alert the home care team when an item is being
overused or supply is too low.
The rising cost of healthcare and the number of diseases worldwide are
transforming the delivery of healthcare from a hospital-centric system to a
person-centric environment (Verma & Sood, 2018). IoT technology provides a
number of personalized solutions that make person-centric healthcare delivery
possible.
IoT technology has the potential of creating many medical applications that
range from remote health monitoring, fitness programs, chronic diseases, and
elderly care (Islam, Kwak, Kabir, Hossain, & Kwak, 2015), help patient monitor
their chronic conditions, recover from injuries or in designing Ambient Assisted
Living (AAL) (Mora, Gil, Azorín, Terol, & Szymanski, 2017). Another area of
healthcare that IoT can potentially address, is compliance with treatment and
medication at home and by healthcare providers (Islam et al., 2015), including
adhering to treatments and medications. Proteus Digital Health, is developing
a pill-size ingestible sensor, that can detect when and how often patients take
their pills (Minteer, 2017). For chronic diseases, an application was designed
for real-time non-invasive glucose level measurements in Diabetes patients
(Istepanian, Hu, Philip, & Sungoor, 2011).
IoT can also capture physiological signs that relate to the human body. Wireless
Body Area Networks (WBAN) allow for the collection and transmission of vital
signs data (such as temperature, blood pressure etc) to enable preventative
care and monitor ones’ well-being (Watts, 2016). Other capabilities of WBAN
include hear pulse, respiration, pulse oximeter (SpO2), electroencephalogram,
pg. 36
electrocardiography (ECG, electromyography, motion detection, cochlear
implant, artificial retina, audio and video) (Adibi, 2012). Another instance of IoT
in Healthcare is an IoT solution for examining human vision for proactive
tackling of vision impairment (Jebadurai & Dinesh Peter, 2018).
Various medical devices, sensors, and diagnostic and imaging devices can be
viewed as smart devices or objects constituting a core part of the IoT (Islam et
al., 2015). The IoT technology also enables interconnection, where for instance,
in healthcare, it can connect health-monitoring devices with emergency
notification systems in real time (Rossman, 2016). With IoT, doctors can play
an active role in their patient’s care by monitoring their health status after being
discharged from the hospital (Rossman, 2016). People are realizing the
importance of healthcare and they seek to obtain medical services
conveniently, making smart healthcare one of the applications of IoT in
healthcare (Wei Li, Cheolwoo Jung, & Jongtae Park, 2018). Lack of timely and
adequate medical information and problems in the management of the care-
giving process are challenges to the effective and efficient implementation of
healthcare services (Kim & Kim, 2018). The features of IoT such as the
seamless connection of sensors and devices can overcome these challenges
by improving information delivery and the care-giving process in healthcare
services (Kim & Kim, 2018).
Table 2.3.2 : IoT Applications. Adapted from (Farahani et al., 2018).
Application Benefits
mHealth Anywhere, anytime access to healthcare
for patients. Doctors can manage and
guide patients when needed
IoT in ambient assisted living Real time monitoring of patients with
chronic diseases and other diseases.
IoT medication Help in detecting adherence to
medication and prevent Adverse Drugs
Reaction.
IoT for individuals with disabilities and
special needs
People with disabilities can access
services to assist and enhance their life.
pg. 37
IoT for smart medical implants Monitor implantable sensors that can
enhance one’s health and treat
diseases.
IoT-based Early Warning Score (EWS) Detect and forecast early signs of
deterioration.
IoT-based anomaly detection Detect early signs of anomalies for faster
treatment.
Table 2.3.2 lists the multiple applications of IoT in healthcare and the benefits
the technology brings to people, healthcare professionals, and the healthcare
domain itself.
IoT technology is successful partly due to the Cloud technology that enables
people to connect, securely, and manage IoT devices, while also generating
insights from data analytics (Patil, 2017). Big Cloud Computing companies like
Amazon (2018), IBM (2018) and Microsoft (2018), all provide IoT services and
solutions for different industries using their cloud offering. Another factor that
made IoT gain popularity is the decrease in the cost of sensors, bandwidth, the
spread of cellular coverage, and the rise of cloud computing that easily
connects devices over the internet (Minteer, 2017). A variety of small
embedded electronic devices in IoT collect and exchange data over the internet
(Wei Li et al., 2018). The IoT market is expanding, such that in Healthcare, IoT
delivering medical devices, systems, software, and services are expected to
grow to a market size of $300 billion by 2022 (Firouzi et al., 2018).
Issues such as aging population mean finding economically viable solutions
that deliver high quality patient centred healthcare services (Qi et al., 2017). IoT
does not only benefit the patients but can potentially assist in easing pressure
off the healthcare systems. IoT enabled technology is changing healthcare from
a conventional hub-based system to a more personalized healthcare system
(Qi et al., 2017). Applications such as remote health monitoring can reduce
costs of healthcare by monitoring non-critical patients at home rather than at
hospital, which reduces the pressure on hospital resources such as doctors and
beds (Baker, Xiang, & Atkinson, 2017). Qi (et al., 2017., ) elaborates that
pg. 38
“Successful utilisation of IoT enabled technology in PHS will enable faster and
safer preventive care, lower overall cost, improved patient-centred practice and
enhanced sustainability. Future IoT enabled PHS will be realised by providing
highly customised access to rich medical information and efficient clinical
decision making to each individual with unobtrusive and successive sensing
and monitoring. (Qi et al., 2017)”.
Reduction in healthcare costs due to IoT is of 2 types. The first is achieved due
to the technological fusion and there is no need to pay separately for the use of
a particular technology (Farahani et al., 2018). Second, patients can monitor
their health status which makes them consult doctors when it is only below the
recommendation (Farahani et al., 2018). From the perspective of healthcare
providers, the IoT has the potential to reduce device downtime through remote
provision. In addition, the IoT can correctly identify optimum times for
replenishing supplies for various devices for their smooth and continuous
operation. Furthermore, IoT provides for the efficient scheduling of limited
resources by ensuring their best use and service of more patients (Islam et al.,
2015).
IoT presents an opportunity to drive change in behaviours that can lead to
healthier lifestyles by connecting pedometers and scales with other tracking
devices (Hoffer, 2016). With the simplicity and versatility IoT provides, and the
technologies involved in tackling diseases, mHealth is another technology that
is readily available using mobile phones that make creating healthcare solutions
possible. Telemedicine is the utilization of medical information transferred
remotely via digital communication to improve or promote the health of patients
and is progressively becoming common practice of medicine in many areas of
the world. The usual model of care for people with cardiovascular diseases
includes regular visits to the doctor and emergency services, involving high
healthcare costs. However, new methods to deliver healthcare using mobile
digital communication devices, mobile health (mHealth), may increase the
number of patients treated while facilitating patient self-management and
saving costs (Supervía & López-Jimenez, 2018).
pg. 39
2.4 mHealth
Mobile technology has changed and will continue to change the life of millions
of people around the globe (Jusoh, 2017). Mobile networks were implemented
as early as 1956 in France, where mobile telephone network devices were
carried in the boot of a car (Le Bodic, 2005). Since then, Mobile technology has
come a long way and the technology has evolved from big bulky devices to
mobile devices, devices that are not physically plugged into a socket or a link
(Jusoh, 2017). The term mobile envisages a number of devices like
smartphones, laptops, tablets, e-book readers, and cameras (Jusoh, 2017).
The most powerful and attractive one is smartphones. It incorporates all
aspects of computing application and wireless communications (Jusoh, 2017).
A number of different emerging and interrelated mobile communication systems
have been adopted and adapted for delivering mHealth, including the
increasing use of (Kreps, 2017):
• Audio and Video Enabled Telemedicine Applications: These applications
include closed circuit telephone and/or televised conferencing systems, with
both data collection tools (often collecting images, or measuring body
temperature, heart rate, and brain activity for diagnostic and monitoring
purposes) and delivery tools (that can remotely control surgical equipment,
and provide treatment instructions) which can be set up in different sites to
provide mediated (virtual) care to patients who may not have easy access to
healthcare delivery systems.
• Telephonic and Internet-based Message Systems: These systems can
provide individuals with health information via SMS text messages, phone
calls, or e-mail via mobile phones.
• Websites and Web-based Portals (intranets): Web portals are available to
healthcare consumers via portable lap-top computers, tablets, and smart
phones.
• Wearable (sometimes implantable) devices: These include devices such as
fitness tracking wristbands and other digital monitoring devices.
• Specialized Application Software Programs (apps): Apps are typically
downloaded onto smart phones, tablets, or laptop computers that provide
pg. 40
health information and support, and are often linked to Internet delivered
resources and services.
Technologies such as mobile and web applications, are becoming an
alternative for people wanting to self-manage their health, generate health data
and gain social support (Arslan, 2016). Benefits of consumer online health-
information seeking (Henriquez-Camacho, Losa, Miranda, & Cheyne, 2014)
are:
• Widespread access to health information in remote areas and vulnerable
populations.
• Equity in access to health information.
• Interactivity with the web contents, in the world of web 2.0 and social
media. Interactivity facilitates interpersonal interaction.
• Tailoring information: interactive health communication offers the potential
for more personalized service and users can select sites, links and specific
messages based on knowledge, educational or language level, need, and
preferences.
The evolution of mobile technology has had an impact on peoples’ lives and the
way we interact with the technology and the people around us (Goggin, 2008).
The technology has allowed people to interact with anyone, anywhere, and at
any time while being reachable (Klemettinen, 2007). This level of interaction
has changed how people stay informed, agreeing on how to do things by
scheduling things, adapting to changes, and maintaining a general feel of each
other’s’ situation (Klemettinen, 2007). Mobile phones have transformed
industries like banking, airlines, transportation, and shopping, by changing the
level of business and customer interaction, and making it mobile living
(Konschak et al., 2013).
Today’s smartphones are a lot more personalized as they store and exchange
personal information, and are more customizable, fitting users’ needs (Boulos,
Brewer, Karimkhani, Buller, & Dellavalle, 2014). The portability and versatility
of mobile phone devices has allowed it to be part of solving different healthcare
pg. 41
challenges in different parts of the world due to low costs, and availability of
mobile devices. The treatment of chronic obstructive pulmonary disease
(COPD) at home, is an example where hospital at home schemes have reduced
the costs of healthcare while maintaining similar effectiveness of care at
hospital and patient safety (Bitsaki et al., 2017).
Mobile technology creates connections between people, information, and
places that allow for designing of health services that link and enable
collaboration among different actors of the healthcare system (Arslan, 2016).
mHealth and the technology known as mhealth technology are extending the
healthcare services beyond physical wardens and doctors. These technologies
are breaking a number of barriers, such as physical barriers (Taylor, 2013), by
extending the services beyond physical places and through remote services
using the technology. Table 2.4.1 illustrates different examples of mHealth for
both clinical and nonclinical applications.
Table 2.4.1: Applications of mHealth. Adapted from (Mirza, Norris, & Stockdale, 2008).
Clinical
mHealth
Applications
Web access to evidence-based databases
Medication alerts using mobile phones
E-prescribing for repeat prescriptions via mobile phones
Telemonitoring to transmit patient results to clinicians Transmission
of test results to patients via SMS messages Online electronic health
records via computer or phone Community nursing contact with
clinical expert advice
Public health and lifestyle messages over mobile phones
Care of at-risk people, e.g. airline pilots, military personnel
Emergency care for accidents, natural disasters
Non-clinical
mHealth
Applications
Efficient workflow via wireless communication
Rapid collection/sharing of current data via mobile phones Optimal
asset utilization, e.g. hospital bed rostering
Patient or asset (e.g. clinical equipment) location using RFID Patient
appointment booking and alerts via wireless e-mail Mobile phone
support for patients and carers
Safety of staff checks with RFID or mobile phones/networks
pg. 42
These technologies have opened the door for new opportunities for people to
use. Chiarini, Ray, Akter, Masella, and Ganz (2013) systematic review of
literature, have identified 4 different solutions that use mHealth systems. Self-
healthcare management, Assisted healthcare, Supervised healthcare, and
continuous monitoring (Chiarini et al., 2013). These solutions are being
delivered via mHealth using the capabilities available on the mobile phone
devices. O'Shea, McGavigan, Clark, Chew, and Ganesan (2017) have
categorized mHealth as follows:
• Video Conferencing
• Text Messaging
• Applications
• Remote Device Monitoring
In Australia, several solutions are available for people wanting to access
healthcare services. An example of a telemedicine in Australia are digital health
services that are becoming more widely available and in many cases are
increasing people’s access to health information (healthdirect, 2018). For
example, the Pregnancy, Birth and Baby service provides video call access to
maternal child health nurses, thus allowing parents to have care delivered to
them without having to leave their home (healthdirect, 2018). In a report
published by PwC (2018), patients expect mHealth to change their healthcare
experience (See Figure 2.4.1). These expectations can result in not only more
health services, but also a reduction in costs.
pg. 43
Figure 2.4.1: Source PwC Report, based on data from Economist Intelligence Unit 2012
Patients expect mhealth application/services to be more convenient for them,
improve the quality of care they receive, and reduce costs associated with
healthcare (costs such as travel to and from the clinic) (PwC, 2018).
The benefits mhealth brings to healthcare include (Gleason, 2015):
• Low cost of equipment
• Patient familiarity with the devices
• Generic interfaces that can be customized for handicapped persons
• Seamless data upload with accurate timestamps and GPS markers built-in
• Existing security systems
• Use of individual equipment identifiers such as the MAC address
• Facilitates record keeping for personalized healthcare
• Automatic user feedback
According to the systematic review by Coorey et al.,4 several risk factors and
lifestyle habits might be modifiable with the utilization of mobile apps, like high
blood pressure, obesity, smoking, dyslipidaemia, and sedentarism, with
improvement of medication adherence, quality of life and psychosocial well-
being (Supervía & López-Jimenez, 2018).
pg. 44
In India, a survey on the use of mHealth for the prevention of CVDs in the Kerala
district has shown that out of 262 participants involved in the survey (Feinberg
et al., 2017):
• 92% of respondents were willing to receive mHealth advice
• 94% favored mobile medication reminders
• 70.3% and 73% preferred voice calls over short messaging services (SMS)
for delivering health information and medication reminders, respectively
• 85.9% would send home recorded information on their blood pressure,
weight, medication use and lifestyle to a doctor or to an accredited social
health activities (AHSA)
• 75.2% trusted the confidentiality of mHealth data, while 77.1% had no
concerns about the privacy of their information.
Among the younger generation, mHealth has been perceived as supporting
young people's self-management for a range of NCDs including asthma,
diabetes, and cancer (Slater, Campbell, Stinson, Burley, & Briggs, 2017). The
self-management is the result of functionalities offered by mHealth technologies
that assist young people in managing their conditions in a number of different
ways, including (Slater et al., 2017):
• Monitoring their health status and symptom triggers via graphical charting
and sign or symptom awareness using self-checks
• Improving their comprehension and understanding of their health condition
• Providing reminders about medication adherence
• Providing ready access to automated tailoring of personal health information
related to the management of their condition(s)
• providing relevant information, support, and reassurance about planning for
emergencies and safety issues through prompting timely communication
with health professionals
Mobile health applications (also knowns as Apps), are becoming more apparent
on Application stores such as Google Play and Apple Store. The range of
applications vary from Apps for medical providers, specialty or disease-specific
apps, medical education and teaching, Apps for patients and the general public
pg. 45
(including health and fitness apps), Text messaging, Telemedicine and
telehealthcare, and smartphone attachments (Boulos et al., 2014). Their
potential means better access to healthcare and more resources for the
management of different diseases.
2.5 Rise of mHealth
The advances in smartphones have made the provisioning of health services
using mobile devices a reality (Adibi, 2013). mHealth, also known as medical
and public health practice supported by mobile devices, such as mobile phone,
patient monitoring devices, personal digital assistants and other wireless
devices (Chow, Ariyarathna, Islam, Thiagalingam, & Redfern, 2016). In a report
published by PwC, mHealth being a disruptive technology is attracting early
adopters who are ill-served by the existing provision or not served at all (PwC,
2018). The two types of patients attracted to such adoption are those with poorly
managed chronic diseases and those who pay more than 30% of their
household income towards healthcare (PwC, 2018). The potential of mobile
phones being used as health devices extends to countries, specifically those
with low-resource settings of low- and middle-income countries where
healthcare infrastructure and services are often insufficient (Chib, van
Velthoven, & Car, 2015).The rise of mHealth is led by existing factors such as
the adoption of mHealth for the aging population, accessibility to health services
self-management, using it for interventions, adherence, health promotion,
seeking information online, and mhealth Apps. These factors are illustrated
below in greater depth.
2.5.1 Aging Population:
The number of aging population is increasing, as WHO (2018) predicts that the
number of people aged 65 or over will be 1.5 billion by 2050, an increase of
8% from 524 million in 2010. An example of this already happening is China as
the country is facing a major challenge caused by the increasing number of the
aging population (Sun, Guo, Wang, & Zeng, 2016). China’s aging population is
estimated to increase at a rate of 5.96 million per year from 2001 to 2020 and
then 6.2 million per year from 2021 to 2050, and is expected to exceed 400
pg. 46
million by 2050, accounting for 30% of its total population (Sun et al., 2016).
One way for mHealth to be used for the aging population is:
• Electronic health records
• e-Prescribing
• Telemedicine
• Consumer health informatics
• Health knowledge management
• mHealth
• Healthcare information systems
• Social media
Mobile technologies such as phones and wireless monitoring devices are
increasingly being used in healthcare and public health practice for
communication, data collection, patient monitoring, and education, and to
facilitate adherence to chronic disease management (Hamine, Gerth-Guyette,
Faulx, Green, & Ginsburg, 2015). mHealth devices can be used in healthcare
to improve service delivery and impact patient outcomes (Hamine et al., 2015).
The Sensors and context-awareness features allow for individualization and
real-time information submission delivery (Hamine et al., 2015). mHealth is
appropriate for the delivery of healthcare services due to the strong attachment
people have to mobile phones as they tend to carry them everywhere, which
opens up opportunities for continuous symptom monitoring and connecting
patients with providers outside of healthcare facilities (Hamine et al., 2015). A
technology also delivering such solutions outside of healthcare facilities is
telemedicine. Telemedicine is the utilization of medical information transferred
remotely via digital communication to improve or promote the health of patients
and is progressively becoming common practice in medicine in many areas of
the world (Supervía & López-Jimenez, 2018). This technology is changing the
way patients interact with doctors. For instance, the usual model of care for
people with cardiovascular diseases includes regular visits to the doctor and
emergency services, which ends up involving high healthcare costs.
technologies such as telemedicine and mHealth, may be able to increase the
number of patients treated while facilitating patient self-management and
pg. 47
saving costs, thus attending to more elderly patients (Supervía & López-
Jimenez, 2018).
2.5.2 Accessibility:
WHO (2018) states that “Understanding health as a human right creates a legal
obligation on states to ensure access to timely, acceptable, and affordable
healthcare of appropriate quality as well as to providing for the underlying
determinants of health, such as safe and potable water, sanitation, food,
housing, health-related information and education, and gender equality”. The
mobility of healthcare services has increased reachability, accessibility and has
the ability to reach more individuals, especially in rural areas (Sezgin, Özkan-
Yildirim, & Yildirim, 2018). The developments of new technologies and web
applications are allowing medical doctors to practice medicine over long
distances and patients have faster access to healthcare services (Bert,
Giacometti, Gualano, & Siliquini, 2013). Technologies like telemedicine and
mHealth are growing and their practical applications can potentiality offer
healthcare services in terms of accessibility of service, cost saving, efficiency,
quality and continuity of care and in clinical practice (Bert et al., 2013). The use
of mHealth in the healthcare domain has made some unique and important
contributions to the delivery of care and the promotion of health as mHealth
applications are used as supplements to traditional channels of health
communication, according to (Kreps, 2017). For example, in the context of
access to healthcare services, mHealth applications have resulted in a timely
delivery of healthcare services and the ability to provide relevant health
information to key audiences wherever they may be via their mobile phones
that they carry with them (Kreps, 2017).
2.5.3 Self-management:
Results from a recent poll in the US showed that the majority of US adults owns
a smartphone, and the interest in mHealth technologies was also reported
amongst the adults, who were keen to adopt mobile fitness technologies
(Dugas et al., 2018). The advantages offered by mHealth technologies are such
that they can be used as interventions due to their widespread accessibility,
pg. 48
cost-effective delivery, and flexibility to content tailoring (Dugas et al., 2018).
The benefits of mHealth provision, have resulted in an increase in the use of
mHealth as a self-management tool for chronic diseases (Dugas et al., 2018).
For instance, CVD diseases result in deaths and disabilities worldwide, with the
number of deaths resulting from CVD expected to reach 22.2 million by 2030
(Baek et al., 2018). 80% of what causes CVD diseases are related to smoking,
drinking, and other unhealthy lifestyle (Baek et al., 2018). To prevent and
reduce these risk factors amongst CVD, mHealth can be used as a self-
managing tool to prevent and manage such chronic diseases (Baek et al.,
2018).
2.5.4 Intervention:
Mobile devices offer unique opportunities for health promotion professionals to
design tailored interventions that make use of innovative technology to reach
populations according to White, Burns, Giglia, and Scott (2016). With the built-
in sensors inside smartphones, physical activity interventions can be delivered
using the built-in pedometers and accelerometers to accurately record exercise
and movement, while nutrition interventions can be delivered using food diaries
and can benefit from non-textual data entry, including the use of photos (White
et al., 2016). Other mHealth apps incorporate strategies such as gamification,
social connectivity and push notifications to reach people as they go about their
daily lives (White et al., 2016). The most common mobile health (mHealth)
approach is a variety of daily and weekly one- or two-way SMS (Short Message
Service) communication interventions that encourage patients to take their
medication (Bardosh, Murray, Khaemba, Smillie, & Lester, 2017). The use of
SMS for interventions offers the opportunity for dissemination of automated,
timely, and target-specific messages, which can be designed to complement or
mirror in-person counselling (Bardosh et al., 2017). For example, messages
offer the opportunity to provide tailored advice, behaviour tracking, goal setting,
encouragement, or personal feedback in different stages of behaviour change
(Bardosh et al., 2017).
pg. 49
2.5.5 Adherence:
Levensky and O’Donohue (2006, p. 3) state that “advancements in healthcare
have yielded numerous effective medical and behavioural health treatments
which, if administered correctly, can help patients live healthier, happier, and
longer. However, all too often the benefits of these treatments are not fully
realized because of patient nonadherence”. The management of chronic
diseases usually require long term care plan, and adherence to chronic disease
management is critical for achieving improved health outcomes, quality of life,
and cost-effective healthcare (Hamine et al., 2015). Poor adherence is a
complex problem that can be divided into intentional (consciously deciding) and
unintentional (forgetting or not being able to use medicines) non-adherence
behaviours (Nousias et al., 2018). People seeking medical care are expected
to change their behaviour (ie: coming to appointments, picking up medications,
agreeing to have assessments and procedures performed) and change in
behaviour in regard to medications (ie: following demanding and complex
medication regimens; making dietary, activity or other lifestyle changes;
enduring sometimes aversive behavioural interventions such as self-monitoring
or exposure) (Levensky & O'Donohue, 2006). However, adherence can be
achieved using mHealth. Asthma and COPD are chronic inflammatory
conditions of the airways affecting over 235 million people worldwide with more
than 30 million living in Europe (Nousias et al., 2018). Inflammatory lung
diseases significantly deteriorate the quality of life for patients and their families
while affecting the overall efficiency of the healthcare system (Nousias et al.,
2018). The adherence of patients to their medication, both in terms of following
the doctor prescription and using the inhaler device correctly, is one of the most
important factors for the effective management of their condition (Nousias et
al., 2018). To assist with the issue of adherence in such disease, a mHealth
system was designed for monitoring medication adherence for obstructive
respiratory diseases that included audio recording of exhalation, inhalation drug
usage, and ambient sound events (Nousias et al., 2018). Several emerging
behaviour theories suggest that SMS- or email-based interventions can provide
timely health messages that can match the level of an individual's motivation
pg. 50
and his or her ability to act, and therefore facilitate behaviour change (Bardosh
et al., 2017).
2.5.6 Health promotion:
Lupton (2012) explains “the use of mhealth in health promotion extends the
temporal nature of health surveillance and allows for further refinements of the
categorising and identifying of ‘risk factors’ and ‘at-risk groups’ that are then
deemed eligible for targeting”. This may be achieved by using the health-related
data that is easily and frequently collected from users’ mobile devices each time
they log on to the relevant app, which becomes a way of monitoring and
measuring individuals’ health-related habits (Lupton, 2012). Promoting health
programs using technologies such as SMS- or email-based have several
advantages over traditional mass media for health promotion and disease
prevention, as they provide opportunities for interactive 2-way communication
and target specific, tailored behaviour change communication (Yepes, Maurer,
Viswanathan, Gedeon, & Bovet, 2016).
2.5.7 Seeking health information online:
Mobile technologies such as phones and wireless monitoring devices are
increasingly being used in healthcare and public health practice for
communication, data collection, patient monitoring, and education, and to
facilitate adherence to chronic disease management (Hamine et al., 2015).
mHealth devices can improve service delivery and impact patient outcomes.
Sensors and context-awareness features allow for individualization and real-
time information submission delivery. Moreover, the strong attachment people
have to mobile phones and the tendency to carry them everywhere, opens up
opportunities for continuous symptom monitoring and connecting patients with
providers outside of healthcare facilities (Hamine et al., 2015).
2.5.8 mHealth Applications (Apps):
The capabilities of smartphones, tablets and other devices have created an
opportunity for a growth in the number of mHealth apps to be developed and
deployed (Wildenbos, Peute, & Jaspers, 2018). In April 2015, according to the
pg. 51
American Heart Association, there were 12,991 apps on iTunes and 1420 apps
on Google Play about weight management, exercise, smoking cessation,
diabetes control, blood pressure management, cholesterol management, and
medication management through mHealth for CVD prevention (Baek et al.,
2018). Nowadays, it is estimated there are 259,000 mHealth apps in the major
app stores from 2016 onwards, which boost innovative mHealth apps to be
used for assisting patients in self-management of diseases and independent
living (Baek et al., 2018). This is particularly important for the older adult patient
population, as risks for functional decline and loss of independence increase
with normal aging and accumulation of chronic diseases, approximately from
the age of 50 onwards (Wildenbos et al., 2018). mHealth apps may offer
medication assistance by prompting alerts, provide self-care advice to patients,
facilitate self-monitoring of various biometrics or educate patients on disease
outcomes (Wildenbos et al., 2018). These mHealth advances align well with the
upcoming interest of older adults to integrate technologies into their own
healthcare (Wildenbos et al., 2018).
These opportunities provided by mHealth technologies, are a gateway for using
mHealth for chronic diseases. However, Chronic diseases do come with their
own challenges and have success factors where mHealth can be introduced as
a solution to help patients with chronic diseases.
2.6 Chronic diseases, challenges and success factors
Chronic diseases have a long painful and suffering period that causes disability
and deaths (Morewitz, 2006). Chronic diseases present enormous challenges
to many patients, their families, many providers, and to the healthcare systems
(Nuovo & Nuovo, 2007). The challenges are due to a number of factors
associated with chronic diseases such as smoking, overweight, obesity, poor
nutrition, sedentary lifestyles, and genetics (Morewitz, 2006). Other factors that
contribute to chronic diseases are race, ethnicity, gender, age, and
socioeconomic inequality (Morewitz, 2006).
pg. 52
According to AIHW (2018), “in 2003, smoking was considered to be responsible
for the greatest disease burden in Australia (7.8% of total burden)”. Chronic
diseases are common amongst Australians as 77% of Australians reported
having one or more long-term health problems with people aged 65 and over
having five or more conditions (Harris & Zwar, 2007). In Europe, chronic
diseases are also a leading cause of deaths as it accounts for 2/3 of all deaths
and consume about 75% of the healthcare costs (Brunner-La Rocca et al.,
2016). In low and middle-income countries, chronic diseases are also leading
to deaths as 80% of deaths are a result of cardiovascular disease and diabetes
mellitus while 90% of deaths attributable to chronic obstructive pulmonary
disease (Beratarrechea et al., 2014). It is estimated that by 2030, 23 million
people will die annually from cardiovascular disease, with the majority of people
(representing 85% of the total figure) residing in low to middle-income countries
(Beratarrechea et al., 2014). Table 2.6.1 lists the different factors associated
with a chronic disease and their definitions. The risk factors include both
behavioural and biomedical health determinants. Some of the NCDs are
preventable and they tend to share the same risk factors (Adeyi, Smith, &
Robles, 2007).
Table 2.6.1: Definition of Risk Factors according to AIHW, 2018
Risk Factor Definition
Daily smoking The smoking of tobacco products on
daily basis
Risky/high risk alcohol consumption: An average daily consumption of more
than 50 mLs for males and more than 25
mLs for females
Physical inactivity Not achieving the recommended
amounts of physical activity of 150
minutes per week over at least five days
Insufficient amounts of fruit Usual consumption of two serves of fruit
per day (few than three servers for those
aged 15-17)
Insufficient amounts of vegetables Usual consumption of fewer than five
servers of vegetables per day (few than
four servers for those aged 15-17)
pg. 53
Fat intake Defined as the usual consumption of
whole milk
Obese Defined as having a body mass index
(BMI) of 30 or more
Large waist circumference A measure of, or greater, than 94
centimetres for men and 80 centimetres
for women
High waist-hip ratio A measure of 1.0 or more for men, and
0.85 or more for women
High blood pressure Based on respondent’s self-reports of
having high blood pressure as a current
and long-term conditions, or currently
taking medication for high blood
pressure
Figure 2.6.1: Number of deaths by risk factors (Source: Our ln World In Data, 2018).
These risk factors are associated with the number of deaths worldwide, which
saw an increase. Some of these risk factors are developed at an early stage of
pg. 54
life (gestation), with some becoming chronic in nature, that results in people
needing more care over a long period of time (Adeyi et al., 2007). The growing
number of the older population, along with health inequalities and poor health
behaviours, mean that the issue of chronic diseases is becoming more
apparent and an escalating problem (Saxton, 2011). The trend in the rising
number of chronic diseases is due to aging, a decline in communicable
diseases and conditions related to childbirth and nutrition, including changing
lifestyles that relate to smoking, drinking, diet and exercise (Adeyi et al., 2007).
Table 2.6.2 lists some of the barriers, current initiatives and possible
enhancements to general practice care for people with chronic disease.
Table 2.6.2: Some barriers, current initiatives and possible enhancements to general practice care for people with chronic disease. Adapted from (Harris & Zwar, 2007).
Barriers and problems Current Initiatives Possible enhancements
Lack of effective
multidisciplinary
team care in general
practice
• Practice nurse
involvement in
chronic disease
management
• Team care
arrangements with
allied health providers
• Access to allied
health through
Medicare or More
Allied Health Services
programs in rural
divisions
• Systems to support
better communication
between general
practice and allied
health
• Infrastructure funding
to provide space and
equipment or other
health professionals
within general
practice
Patient understanding of
self-management and
adherence to
management plan
• Sharing Health Care
Initiative
• More available local
self-management
programs
• Involvement of
practice staff in
delivering self-
management
education
pg. 55
• Better feedback from
self-management
programs to general
practitioners
Care that does not meet
evidence-based
guidelines
• Guidelines
disseminated by =
National Health and
Medical Research
Council., Royal
Australian College of
General Practitioners,
Diabetes Australia,
National Heart
Foundation
• Better integration of
guidelines into
structure of practice
information systems
Inability of most clinical
information systems to
provide effective clinical
audit of quality of care
• Templates for Care
Plans
• Support for disease
registers through
Practice Incentives
Program.
• Division and National
Primary Care
Collaboratives
collaborative
programs
• Greater capacity for
audit incorporated
into practice systems
People who suffer from a chronic disease need to adjust to the requirements of
the illness, and adhere to the treatments (Morewitz, 2006). A number of
strategies exist for managing chronic diseases, and prevention is one of the
methods of reducing cases of chronic diseases (Estrin & Sim, 2010). Smoking
is one of the contributing factors of chronic diseases. Reduction in tobacco
smoking rates has reduced the number of males dying from lung cancers
(AIHW, 2018). Similarly, the screening for cervical cancers has reduced the
number of deaths that were caused by cancer (AIHW, 2018).
pg. 56
“Doctors alone cannot solve chronic diseases, but have to work with patients to
manage chronic diseases together so that the disease does not lead to painful
complications like amputations or blindness” as stated by Stuckler and Siegle
(2011, p. 124). A chronic disease care model for preventing chronic diseases
and working with patients, is one that focuses on (Stuckler & Siegel, 2011):
1. Continuity of care
2. Preventing unnecessary hospitalisation
3. Coordinating and integrating care services
4. Empowering patients to know about and manage their conditions
5. Joint decision-making between doctors and patients
A number of methods exist for managing chronic diseases and preventing
incidents of chronic diseases such education, prevention, and support. Chronic
diseases can be treated and managed in different ways. An example of this is
diabetes, as change in lifestyle is seen as a medication for the treatment of
diabetes (Nuovo & Nuovo, 2007). Physical activity also leads to the prevention
and delay in contracting NCD diseases (Phillips, Cadmus-Bertram, Rosenberg,
Buman, & Lynch, 2018). A way of enabling these prevention programs is using
technology.
Technology can have a great impact on chronic diseases and may reduce the
pressure on healthcare systems. In an Australian telehealth study, telehealth
technology monitored patients and the results show (Jayasena et al., 2016):
• Reduction of health services utilization observed towards the later part of
the trial (after 8-9 months) by reduced number of GP visits, specialist
consultations, laboratory tests and procedures performed
• Reduction in primary healthcare costs were demonstrated by the reduction
of Medicare benefit costs by more than AUD600/year/patient and
medication dispensing costs by more than $300/year/patient at end of trial
• Reduction in hospital length of stay by more than 7 days
• Reduction in rate of hospitalisations by 1 less event
• There were improved Quality of life metrics/Human factors/Usability and
acceptability to patients and clinicians
pg. 57
This can be achieved through self-empowerment of patients, which is by
emphasizing the role patients play in the active management of their own
healthcare (Nuovo & Nuovo, 2007). Self-empowerment is an example of the
shift happening in the healthcare model which regards patients as passive
recipients of care, to a model which empowers patients with knowledge and
skills to manage their health (Saxton, 2011). Self-management can be achieved
through mHealth and other technologies that empower patients (Baek et al.,
2018).
With the capabilities and functionalities mHealth has, the technology can be
used for dealing with a variety of chronic diseases.
2.7 mHealth for chronic diseases
According to Ardies (2014, p. 1) “Chronic diseases comprise a major source of
mortality from disease for the world and are by far the greatest source of
mortality among the high-income countries. In the United States, chronic
diseases account for about 70% of all deaths and 75% of healthcare costs and
their prevention is extremely important in terms of alleviating personal suffering
and reducing economic impact, for individuals and for the nation”.
Chronic diseases can range from mild conditions such as short- or long-
sightedness, dental decay and minor hearing loss, to debilitating arthritis and
low back pain, and to life-threatening heart disease and cancers (AIHW, 2018).
Chronic diseases are often long-term conditions that cannot be cured, which
result in people needing long term management of the diseases (AIHW, 2018).
Once present, chronic diseases often persist throughout life, although they are
not always the cause of death. Examples of chronic diseases include:
• Cardiovascular conditions (such as coronary heart disease and stroke)
• Cancers (such as lung and colorectal cancer)
• Many mental disorders (such as depression)
• Diabetes
• Many respiratory diseases (including asthma and COPD)
pg. 58
• Musculoskeletal diseases (arthritis and osteoporosis)
• Chronic kidney disease
• Oral diseases
The most noticeable forms of chronic conditions are Cardiovascular Diseases
(CVD), cancer, chronic respiratory disease (CRD), and diabetes (Bovell-
Benjamin, 2016). WHO (2018) has defined the chronic diseases and illustrated
the burden chronic diseases have on healthcare systems and on world
economies. The following elaborates on chronic diseases:
• Cardiovascular Diseases (CVDs): Are a group of disorders of the heart and
blood vessels including: coronary diseases, cerebrovascular diseases,
peripheral arterial diseases, rheumatic heart diseases, congenital heart
disease and deep vein thrombosis and pulmonary embolism. CVDs have a
huge impact on morbidity as they are ranked as the number 1 cause of
death. In 2015, 17.7 million people died from CVDs, representing 31% of all
global deaths. Some of the risk factors associated with CVDs are unhealthy
diet, physical inactivity, tobacco use and harmful use of alcohol.
• Cancer: Cancer is a generic term for a large group of diseases that can
affect any part of the body. Other similar terms of cancer are malignant
tumors and neoplasms. Cancer is known for the fast creation of abnormal
cells that grow beyond their usual boundaries, and which are able to
penetrate adjoining parts of the body and spread to other organs, which is
known as metastasizing. Metastases are a major cause of death. Cancer is
a leading cause of deaths worldwide accounting for 8.8 million deaths in
2015. The various types of cancers include lung (accounting for 1.68 million
deaths), liver (accounting for 788,00 deaths), Colorectal (accounting for 744
000 deaths), Stomach (accounting for 754 000 deaths), and Breast
(accounting for 571 000 deaths). Risk factors associated with cancer are
tobacco use, alcohol use, unhealthy diet, and physical inactivity.
pg. 59
• Chronic Respiratory Diseases (CRD): Chronic Respiratory Disease (CRD)
also known as Chronic Obstructive Pulmonary Disease (COPD) is a lung
disease that is known by the persistent reduction of airflow. The primary
cause of COPD is tobacco (including secondhand or passive exposure).
Others include indoor air pollution, outdoor air pollution, occupational dusts
and chemicals, and frequent lower respiratory infections during childhood.
It is estimated that 3.17 million deaths worldwide were caused by COPD in
2015. In 2016, a study by The Global Burden of Disease has reported the
widespread of COPD to be 251 million cases worldwide.
• Diabetes: Diabetes is a chronic disease that occurs when the pancreas
does not produce enough insulin or when the body cannot effectively use
the insulin it produces. Insulin is a hormone that regulates blood sugar.
Diabetes include Type 1, Type 2 and Gestational diabetes.
Other diseases such as mental illness, vision and hearing loss, and digestive
diseases such as cirrhosis are also considered chronic diseases (Bovell-
Benjamin, 2016). With the listed risk factors associated with the chronic
diseases, diet and physical activity can potentially reduce their burden by
developing preventive methods that focus on diet and physical activity (Ardies,
2014). Health inequalities are stemming from chronic conditions in low-middle
income countries, resulting in the urgent need to implement more effective and
cost-effective interventions (Beratarrechea et al., 2014). Interventions to
promote and improve patient empowerment are important components for
effective self-care behaviour, risk factor control for chronic diseases, can
increase adherence to medical treatments and preventive strategies (Supervía
& López-Jimenez, 2018). Deaths caused by chronic diseases can be prevented
through counselling, risk factor modification, and medication adherence
(Beratarrechea et al., 2014). An example of risk factor modification using
mHealth is from a trial in New Zealand aimed at ceasing smoking, where text
messaging was used as an intervention method for CVDs (Chow et al., 2016).
The study showed that those receiving text-message intervention were almost
twice as likely to quit smoking (Chow et al., 2016). Table 2.7.1 is an example of
pg. 60
tailored text-messages developed for the self-management of chronic diseases
(Dobson et al., 2015).
Table 2.7.1: SMS4BG Modules. Relevant Sections Adapted from (Dobson et al., 2015)
Module Description Participants Example text
message
Smoking cessation 1 message per
month
encouraging
participants to
consider quitting
smoking and
providing details of
services for
support.
All participants
who register as
smokers.
“SMS4BG: Good
management of
your diabetes and
your future health
includes not
smoking, call
Quitline on 0800
778 778 for
support.”
Lifestyle behaviour Up to 4 messages
per week
encouraging
participants to set
a lifestyle goal and
supporting them to
work toward this
goal. Participants
can receive one of
these modules for
3 months.
The three lifestyle
modules are: (1)
exercise, (2)
healthy eating,
and (3) stress and
mood.
Available to all
participants.
(1) “SMS4BG: Hi
[name]. If you are
finding it tough to
keep up your
exercise think
about why good
management of
your diabetes is
important to you.”
(2) “SMS4BG:
Healthy eating is
an important part
of your diabetes
treatment and it
will help you in
controlling your
blood glucose
levels.”
(3) “SMS4BG:
Make sure you
have fun activities
scheduled
pg. 61
regularly. Doing
something
enjoyable helps
reduce stress &
improves mood.”
Blood glucose
monitoring
Reminders to test
blood glucose,
sent at a
frequency selected
by the participant
(up to 4 per day),
for which they are
encouraged to
reply by text
message with their
blood sugar
readings.
Available to all
participants
required to monitor
their blood
glucose.
“SMS4BG: Hi
[name]. Just a
reminder it is time
to check your
blood glucose.
Reply with the
result.”
If valid response
received,
“SMS4BG: Thank
you for sending
your result.”
The effectiveness of mobile phone text messaging was also related to weight
loss, physical activity, lowering of blood pressure and management of diabetes
(Chow et al., 2016). mHealth has the functionalities that can deliver health
services to people and patients with chronic diseases. An application of this is
mHealth for diabetes.
2.8 mHealth for Diabetes
Diabetes is a long term, non-curable serious diseases that is based on
hormonal conditions where the body is unable to use the energy from the food
(Leslie, Lansang, & Coppack, 2012). Shaw (2012) states that there are 2 main
types of diabetes:
1. Type 1, where the autoimmune system destroys the insulin-producing cells
in the pancreas
2. Type 2, which is a common form and is caused by the reduced production
of insulin and an inability of the body tissues to respond fully to insulin
pg. 62
Hyperglycaemia, also known as high level of blood sugar, is a common effect
of uncontrolled diabetes and it can affect body parts over a period of time,
including the nerve and blood vessels (WHO, 2018).
Diabetes causes complications that affect different parts of the human body that
include feet, eyes, kidneys, and cardiovascular health (Shaw, 2012). The
number of deaths according to WHO (2018), show that in 2014, 8.5% of adults
aged 18 years and old had diabetes, 1.6 million deaths in 2015 from diabetes
being the direct cause, and 2.2 million deaths in 2012 due to high blood glucose.
Diabetes is predicted to be the seventh leading cause of deaths in 2030 (WHO,
2018). Diabetes does not only cause deaths but is a financial burden on families
and individuals suffering from it (Patel, Resnicow, Lang, Kraus, & Heisler,
2018).
WHO (2018) explains that “diabetes affects economy on a global scale as a
result of direct medical costs, indirect costs associated with productivity loss,
premature mortality, and the negative impact of diabetes on nation’s gross
domestic product (GDP)”. The costs of diabetes are associated directly with the
expenditure for preventing and treating it and its complications, including
outpatient and emergency care, inpatient hospital care, medications and
medical supplies such as injection devices and self-monitoring consumables
and long-term care (WHO,2018). The cost of diabetes in Australia is $6 billion
for type 2 diabetes, and $570 million for type 1 diabetes, which includes
healthcare costs, the cost of carers and Commonwealth government subsidies
(Shaw, 2012).
Managing diabetes can be quite complex and requires a lot of effort to ensure
the blood glucose level remains within the desirable threshold of healthy levels.
Although diabetes is classified as a non-communicable disease and it is not
curable, it can be controlled using medications (Saxton, 2011). The prevention
of diabetes can be achieved by changing the lifestyles and through the early
diagnosis and treatment of the disease (WHO, 2018). The changes in lifestyles
include healthy diet, regular physical activity, maintaining a normal body weight
and avoiding tobacco use (WHO, 2018).
pg. 63
Whettem (2012, p. 8) illustrate that “Good diabetes management involves
partnership between healthcare providers and individuals with diabetes actively
managing their condition themselves on a day-to-day basis, a role which may
extend to family, friends and carers/ support workers in the case of children and
young people, older people and people with learning disabilities or mental
health problems”. Good management of diabetes involves people obtaining and
developing knowledge, skills and confidence required to ensure the conditions
are managed effectively (Whettem, 2012). With the global burden of diabetes,
access to technology can enable a better management of diabetes that includes
all activities aforementioned that are delivered using mHealth technology
(Wickramasinghe, Troshani, & Goldberg, 2010). mHealth for diabetes has
shown a positive impact on the management of diabetes. It has improved
glucose management for people and assisted in maintaining lifestyle changes
(Georgsson & Staggers, 2016).
The nature of diabetes management makes technology an appealing solution
for providing patients with support as patients routinely use glucometers,
continuous glucose monitoring (CGM) devices, and insulin pumps
(Modzelewski, Stockman, & Steenkamp, 2018). Dulce Digital is an example of
where SMS based intervention was designed to support middle aged females
born in Mexico (Fortmann et al., 2017). In a systematic review (See Table 2.8.1)
done by Youfa et al. (2017), a number of interventions exist in the literature that
highlight the role of different mHealth technologies for the treatment of diabetes.
Table 2.8.1: Details of mobile application intervention for chronic disease management. Adapted from (Youfa et al., 2017)
Intervention Control
Treatment
Key Outcome(s) Conclusions
Daily tailored
text messages
to enhance
diabetic self-
care behaviour
Standard
diabetes self-
management
1) HbA1c levels changed
but not significantly (P =
0.1534). 2) Self-efficacy
scores improved
significantly (P = 0.008).
Text messages may
have a positive
impact on some
clinical outcomes and
self-efficacy in
pg. 64
patients with type 2
diabetes.
Personalized
messaging in
response to
blood glucose
values,
diabetes
medications,
and lifestyle
behaviours by
mobile phone
Usual care 1) HbA1c decreased
over 12 mo (P <
0.001). 2) Appreciable
differences in
depression, diabetes
symptoms, BP, and lipid
concentrations were not
observed between
groups (P > 0.05).
Telephone
monitoring of
PA and a low-
calorie, low–
glycemic index
diet
Conventional
low-fat diet
and standard
care
Weight loss, glucose,
HbA1c, and antidiabetic
drug intake were
reduced (P < 0.001).
Telephone
monitoring can
effectively lower body
weight, HbA1c, and
antidiabetic drug use
in patients with type 2
diabetes.
Self-care
video
messages by
cell phone
Standard
diabetes care
Decline in HbA1C over
12 mo (P < 0.002).
One-way, mobile
phone–based video
messages about
diabetes self-care
can improve HbA1C.
A short
message
every week on
diet, exercise,
taking diabetic
medication,
self-monitoring
of blood
glucose, and
stress
management
Counseling
appointments
via telephone
Between control and
intervention groups,
there was no significant
difference in HbA1C (P =
0.489), diet adherence
(P = 0.438), or physical
exercise (P = 0.327), but
medication adherence
improved in the control
group (P = 0.034).
SMSs of cellular
phones may improve
adherence to
diabetes therapeutic
regimen in patients
with type 2 diabetes.
pg. 65
Remote
patient
reporting and
an automated
telephone
feedback
system
Standard of
care
Significant reduction in
HbA1c (P < 0.03) and
body weight (P < 0.03).
Automated feedback
improves patient
outcomes in patients
with type 2 diabetes.
Telemetry
device at
home to
transmit
glucose levels,
BP readings,
and weight
measurements
to the diabetes
care manager
weekly
Standard
care
1) Self-efficacy (P =
0.319), fructosamine (P =
0.881), and HbA1c (P =
0.310) had no significant
changes between
groups. 2) LDL tended to
decrease more in the
intervention group (P =
0.045).
The use of a
telemetric device to
download information
for care management
was associated with
improved control of
LDL-C.
For the treatment of a health issue to be successful, an active participation by
an informed patient is required (LeRouge & Wickramasinghe, 2013). An
instance where this is valid, is in the case of lifestyle issues related to chronic
diseases, such as diabetes, which cannot be successfully treated without
changing the patient’s personal behaviour and habits (LeRouge &
Wickramasinghe, 2013). In the past, a number of information and
communication technologies (eg, Internet, telephone, mobile phone, and
Bluetooth-connected blood glucose meters) were used to track and transmit
blood glucose results (Cafazzo, Casselman, Hamming, Katzman, & Palmert,
2012). Many of these interventions were seen as electronic logbook
applications to collect, store, and display blood glucose readings (Cafazzo et
al., 2012). Some of these technologies included prompts to cue patients with
diabetes about taking actions and modifying the treatment protocolCafazzo et
al., 2012).
pg. 66
In a systematic review conducted by Kitsiou, Paré, Jaana, and Gerber (2017),
it was found that a variety of mHealth interventions were included in the
systematic reviews which used a variety of technological innovations, including
text messaging (SMS), mobile applications, Bluetooth-enabled glucose meters,
as well as secure websites/web-portals that could be accessed through the
participants’ mobile device for data entry and patient support (Kitsiou et al.,
2017). The review looked at how technology is used for diabetes and it found
that (Kitsiou et al., 2017):
• The most prevalent method for sending blood glucose measurements
and/or receiving self-management support (e.g. encouragement, education,
reminders, and recommendations) involved the use of text messages (SMS)
via mobile phones (34 out of 53 studies).
• Twelve studies involved the use of web-portals that patients accessed
through their cell phones to enter diabetes self-care data and receive
feedback.
• In 19 studies, data entry and transmission of glucose data were performed
automatically via Bluetooth-enabled devices or other connectivity methods
(e.g. infrared transmission or attachment of the glucose device to the mobile
phone). In the remaining studies data entry was performed manually.
• The majority of studies included in the systematic reviews (38 out of 53)
encouraged self-monitoring of blood glucose combined with other
intervention components, such as support of medication adjustment and
reinforcement of lifestyle changes.
• Half of the studies (26 out of 53) provided educational support to patients
via SMS (15 studies) and/or the Internet (11 studies). In 28 of 53 trials
patient data were transmitted daily or more often.
Table 2.8.2: Examples of tailored text-messaging for self-management support for Blood Glucose (SMS4BG) (Dobson et al., 2015)
Module Description Participants Example text message
Core 2 messages per week
providing general
motivation and support
for diabetes
All (1) “SMS4BG: Kia ora.
Control of your glucose levels
involves eating the right kai,
exercise & taking your
pg. 67
management.
Available in two
versions: (1) Māori, and
(2) non-Māori.
medication. Your whanau,
doctor & nurse can help you.”
(2) “SMS4BG: There is no
quick fix to diabetes but with
good management it will have
less impact on your life and
leave you more time to do the
things you enjoy.”
Insulin 1 educational text
message per week on
insulin management for
patients receiving
insulin.
Available to
participants
prescribed
insulin.
“SMS4BG: Unopened insulin
should be kept in the fridge.
Don’t use insulin that has
changed color, lumpy,
expired, cracked or leaking,
has been frozen or too hot.”
Young
adult
1 message per week
around managing
diabetes in the context
of work/school and
social situations.
Available to
participants
aged 16-24.
“SMS4BG: It’s important not
to ignore a hypo. No one likes
to be embarrassed, but
ignoring a hypo can make you
feel worse & can be more
embarrassing.”
Another study developed tailored text messages (See Table 2.8.2) for assisting
people with self-management of blood glucose (Dobson et al., 2015). This
illustrates how mHealth can be used to assist people in managing their
diabetes.
The introduction of technologies in healthcare has changed the interaction
between patients and clinicians. Patients using mHealth technologies for the
management of their chronic conditions, have taken the patient-clinician
interaction to a new level with the aid of smartphones.
2.9 Patient – Clinician Interaction
People in need of care often interact with different stakeholders of the
healthcare system. This interaction is often referred to as patient-clinician
interaction. The interaction, also known as patient-practitioner interaction, is
“the relationship between practitioner and patient in the context of the
pg. 68
consultation” (Dambha-Miller, Silarova, Kinmonth, & Griffin, 2017, p. 1) Kelly-
Irving et al. (2009, p. 1) explains that “the interaction between patients and their
general practitioner is important for the efficiency and usage of health services,
and the nature and quality of the relationship affect communication, medical
advice, satisfaction and diagnosis”.
The interactions often include communication, empathy, interpersonal skills,
listening, and mutual decision-making (Dambha-Miller et al., 2017). When
dealing with a chronic disease, these interactions provide an opportunity for
health professionals to influence patients’ behaviour (Dambha-Miller et al.,
2017). Health professionals can positively influence healthcare decisions,
encourage self-management or enablement, and support patients with long-
term health changes during the long course of diabetes (Dambha-Miller et al.,
2017). These interactions can be decisive and have a positive outcome on
patients’ behaviour as early evidence suggests that better patient experiences
with practitioners are associated with lower cardiovascular (CV) risk factors
such as blood pressure, glycated haemoglobin (HbA1c), and cholesterol, which
in turn could lead to a delay in diabetes progression (Dambha-Miller et al.,
2017). Clinicians relationships with patients can be as healing as the treatment
they recommend, as the therapeutic effect from the interaction can have greater
effect than the pharmacological effects (Halpert, 2018). Patients interacting with
medical professionals often have perceptions about the care they receive.
Halpert (2018, p. 1) describes the effects of the patient-provider interaction on
placebo (inert substance that provokes perceived benefits) or nocebo (inert
substance that causes perceived harm) provide strong evidence that the
context of the patient-provider encounter impacts clinical outcomes.
People who do not participate in diagnosing or preventing diseases belong to
key groups unaware of the risk of severe or life-threatening diseases and do
not perceive the benefits of behaviour change outweighing potential barriers or
other negative aspects of recommended actions. (Schiavo, 2013). A Health
belief model (HBM) was established for the purpose of explaining why people
do not participate or undertake proactive measures in managing their health.
HBM has the following components (Schiavo, 2013):
pg. 69
• Perceived susceptibility: The individual’s perception of whether he or she is
at risk for contracting a specific illness or health problem
• Perceived severity: The subjective feeling whether the specific illness or
health problem can be severe (for example, permanently impairing physical
or mental functions) or is life threatening, and therefore worthy of one’s
attention
• Perceived benefits: The individual’s perceptions of the advantages of
adopting recommended actions that would eventually reduce the risk of
disease severity, morbidity, and mortality
• Perceived barriers: The individual’s perceptions of the costs of and
obstacles to adopting recommended actions (includes economic costs as
well as other kinds of lifestyle sacrifices)
• Cues to action: Public or social events that can signal the importance of
taking action (for example, a neighbour who is diagnosed with the same
disease or a mass media campaign)
• Self-efficacy: The individual’s confidence in his or her ability to perform and
sustain the recommended behaviour with little or no help from others
Complying with instructions provided by doctors is crucial to the process of
treatment, and effective therapy (Heszen-Klemens & Lapińska, 1984). Patients
who do not follow the instructions, often do not understand or remember the
insutrctions due to emotional reactions to sickness or incapable of
understanding the instructions (Heszen-Klemens & Lapińska, 1984). Table
2.9.1 presents the essential elements of effective patient-clinician
communication inside hospitals.
Table 2.9.1: Patient-clinician communication in hospitals (Source: safetiyandquality.gov.au, 2018)
Element Purpose Outcome
Fostering
relationships
• To build rapport, trust
and good relationships
• Improved satisfaction and
experience with the health
service
• Trust in the health service
pg. 70
• A decrease in healthcare
provider stress and burnout
Two-way
exchange of
information
• To ensure accurate
diagnosis and
interpretation of
symptoms
• To engage with a person
to gather relevant
information
• To share meaningful
information in a
comprehensive way
• To check that a person
understands the
information provided
• Increased diagnostic
effectiveness and improved
health outcomes
• Less medical errors
• Increased and shared
understanding of a person’s
care needs and preferences
• Decision making based on
complete and accurate
information
• Improved partnerships
between people and their
healthcare providers
Conveying
empathy
• To build rapport, trust
and good relationships
• To deliver quality
healthcare
• To acknowledge and
treat the patient as a
person
• Improved satisfaction and
experience with the health
service
• Improved partnerships
between people and their
healthcare providers
Engaging
patients in
decision-
making and
care planning
• To reach agreement on
problems and plans
• To facilitate self-
management
• To recognise that the
person receiving care as
an important role in co-
producing their care
• To ensure that decisions
are appropriate, realistic
and reflects the person’s
preferences and goals
for their care
• Improved health outcomes
• People having a better
understanding of their care
plan and treatment
• Improved adherence to care
and treatment
• Increased ability for a person
to self-manage their care
• Improved satisfaction by a
person of the decisions made
about their care
pg. 71
Managing
uncertainty
and complexity
• To manage a person’s
expectations
• To keep the person
informed
• To ensure that care is
appropriate to the
changing needs of a
person receiving care
• To ensure that critical
information is not lost
when care is transferred
between teams
• To ensure new
information is
communicated and
considered
• A reduction in the potential
distress, anxiety or confusion
arising from changes to care
• Improved satisfaction and
experience with the health
service
• Treatment and care matching
the care needs of the person
• Improved health outcomes
In chronic disease, clinicians need to support patients as they engage in self-
management and adhere to medications. In diabetes care, clinician empathy
has been associated with objective measures of disease management such as
blood sugar control and fewer complications of diabetes (Flickinger, Saha et al.
2016). Empathic communication has also been shown to increase patients’
cancer-related self-efficacy and sense of control (De Vries et al., 2014).
Flickinger et al. (2016, p. 221) states that “when primary care patients rate their
clinicians as having higher empathy, they demonstrate better adherence to
recommended treatment. Showing empathy plays a role in patient adherence
as it is develops interpersonal trust and a therapeutic partnership between
clinicians and patients (Flickinger, Saha et al. 2016). Flickinger et al. (2016, p.
221) also states that “it may also be that clinicians who exhibit more empathic
communication are able to create more effective partnerships with patients,
providing patients with the understanding and confidence necessary to take
active roles in disease management and thus, achieving more favourable
disease outcomes”.
pg. 72
According to De Vries et al (2014, p. 1) “Effective communication allows the
clinician to assess, inform, and support the patient and has been associated
with positive patient outcomes such as physical and emotional wellbeing, pain
control, adherence to treatment, accuracy and completeness of assessment of
symptoms and side-effects, patient satisfaction, information recall, and
psychological adjustment. Studies investigating factors that influence”.
Communication in healthcare between patients and clinicians are now being
transformed using IT (Weiner, 2012). Such transformations include face-to-face
patient/doctor contact becoming less desirable, and the exchanges between
consumers and providers will increasingly be facilitated by electronic devices
(Weiner, 2012). The presence of online information, such as Online Health
Communities (OHC), is changing the way patients seek care (Atanasova,
Kamin, & Petrič, 2018). Consulations done through OHCs provide patients with
many benefits including a convenient, accessible, geographically independent
and reliable source for information emotional support (Atanasova et al., 2018).
As OHCs are internet-based platforms, they connect different groups of
individuals with similar health-related interests, which in turn become important
venues for connecting people with similar health conditions and sometimes
access to health professionals (Atanasova et al., 2018). OHCs are divided into
2 types (Atanasova et al., 2018):
1. OHCs, which are mainly devoted to peer support groups and are usually
referred to as online support groups.
2. The second type, which are most commonly associated with the term of
OHC, refers to online platforms that include both patients and health
professional moderators who are usually healthcare professionals or
doctors.
With the information available online and through communities, they become a
key enabler in decision making. The exchange of information is a pre-requisite
of effective decision making and is in turn influenced by the quality of the
relationship existing between doctor and patient as described by (Bigi, 2016).
The exchange of information, data produced by electronic devices, and the data
pg. 73
generated during these interactions, produce a huge amount of data, and is
often referred to as big data. Big data is another source of technology used in
healthcare to assist in diagnosing patients and diseases, produce information
from the analysis of large datasets and generate value from the collected data.
2.10 Big data for Health
Devices such as mobile phones, wearable technology, tiny sensors, social
media, and other modern technologies, are producing large amount of data that
led to the phenomenon of big data (Runciman & Gordon, 2014). The huge
volume of data available in different industries and domains, is one of the
characteristics of big data. De Mauro, Greco, and Grimaldi (2015) state that,
“Big Data represents the Information assets characterized by such High
Volume, Velocity and Variety to require specific Technology and Analytical
Methods for its transforming into Value”. This definition took into consideration
the 4 themes of Information, Technology, Methods, and Impact, which are
always found when big data is defined (De Mauro et al., 2015).
At the time big data emerged, the 3 V’s of big data were Volume, Velocity and
Variety (Trifu & Ivan, 2014). However, these have been expanded and now
include Veracity, Variability, and Value (Sebaa, Chikh, Nouicer, & Tari, 2018).
The definition of the 6 V’s are (Sebaa et al., 2018):
• Volume: High volume of data
• Velocity: Rate of data generation and speed of processing
• Variety: The diversity of types of data
• Veracity: The degree of reliability and quality of sources of data
• Variability: The variation in rate of data flow
• Value: Value generated based on the analysis of data
Examples of some of the V’s in healthcare are (Lin, Ye, Wang, & Wu, 2018):
• Volume: A common statistic from EMC (now known as Dell EMC) is that 4.4
zettabytes of data existed globally in 2013, which is predicted to increase to
44 zettabytes by 2020.
• Velocity: Health monitoring data is being generated every second
pg. 74
• Variety: The medical domain contains a number of big data sources such
as the digital medical record, MRI, CT, health monitoring data, and genome
data.
• Veracity: Medical data might be incomplete, biased, or even filled with noise
Big data differs from Business Intelligence (BI) as its features are different from
Traditional BI as shown in Table 2.10.1 (Trifu & Ivan, 2014).
Table 2.10.1: Business Intelligence vs Big Data, Adapted from (Trifu & Ivan, 2014).
Traditional BI Reporting Big Data Analysis
Reporting tool like cognos, SAS, SSIS,
SSAS
Visualization tool like QlikView or
Tableau
Sample data or specific historical data Huge volume of data
Data from the enterprise Data from external sources like social
media apart from enterprise data
Based on statistics Based on statistics and social sentiment
analysis or other data sources
Data warehouse and data mart OLTP, real-time as well as offline data
Sequential Computing Parallel Computing using multiple
machines
Query languages SQL, TSQL Scripting Languages Java script,
Python, Ruby and SQL
Specific type of data : txt, xml, xls, etc Multiple kinds of data : pictures, sounds,
text, map coordinates
The Healthcare domain contains a big volume of data and information
(Koumpouros, 2014). One of the many sources of big volume of data in
healthcare is data that is generated by patients during their interaction with
healthcare, such as hospital visits (Koumpouros, 2014). Medical professionals
also collect data while interacting with patients and use them to achieve a
desired outcome (Koumpouros, 2014). Other sources of big data in healthcare
include data generated by the use of the Electronic Health Records (EHR),
personalized medicine, and administrative data (van Loggerenberg,
Vorovchenko, & Amirian, 2017).
pg. 75
Preventative care is understood to be better than cure, as information from the
analysis of big data is put into practice (Marr, 2015). The collections of cancer
data from patients have resulted in large datasets requiring method from big
data to discover new knowledge and aid in the analysis of data (Hinkson et al.,
2017). Big data is changing healthcare as it delivers new insights through the
analysis of data. Search engines and social media can provide insights into
people’s reaction and monitor their conditions of epidemic diseases (Huang et
al., 2015). Decision support systems, are another example where information
from big data can be used to help individuals choose their medicine and assist
health professionals in selecting the treatment for patients (González-Ferrer,
Seara, Cháfer, & Mayol, 2018).
The information relating to patients’ care and conditions is derived from the
analysis of big data. A single patient can ultimately produce around 1.2
Gigabytes of structured data consisting of patient record (Marconi & Lehmann,
2014). Data from a variety of different sources such as hospital information,
medical sensors, medical devices, websites and medical applications consume
storage capacity in the masses of terabytes and petabytes of storage (Sebaa
et al., 2018). Big data has the promise of answering some of the challenging
questions in disciplines such as science, technology, healthcare, and business,
that are yet to be addressed (Tomar, Chaudhari, Bhadoria, & Deka, 2016).
Big data can address many of the problems in health sciences such as
recommendation system in healthcare, Internet based epidemic surveillance,
sensor based health conditions and food safety monitoring, Genome-Wide
Association Studies (GWAS) and expression Quantitative Trait Loci (eQTL),
inferring air quality using big data and metabolomics and ionomics for
nutritionists (Gao, Chen, Tang, Zhang, & Li, 2016).
Some of the companies delivering big data in healthcare are IBM. IBM Watson
technology uses big data for healthcare solutions. IBM Watson be used in
multiple areas of healthcare (IBM, 2018):
pg. 76
• Drug Discovery: Used for helping researchers identify novel drug targets
and new indications for existing drugs and uncovering new connections and
developing new treatments ahead of the competition.
• Social Program Management: Supporting agencies in their work to deliver
health and human services that enable citizens to meet their maximum
potential while protecting the most vulnerable populations.
• Oncology: Spend less time searching literature and more time caring for
patients. For example, IBM Watson can provide clinicians with evidence-
based treatment options based on expert training by Memorial Sloan
Kettering physicians.
• Care Management: Use personalized care plans, automated care
management workflows, and integrated patient engagement capabilities to
help create more informed action plans.
The role of Big Data in healthcare has already been seen in various
applications. Precision Public Health, is a term defined as “improving the ability
to prevent disease, promote health, and reduce health disparities in populations
by applying emerging methods and technologies for measuring disease,
pathogens, exposures, behaviours, and susceptibility in populations; and
developing policies and targeted implementation programs to improve health”
(Dolley, 2018). Precision public health uses big data and its enabling
technologies to accomplish a previously impossible level of targeting or speed
(Dolley, 2018). Big data has been used in Precision Public Health for (Dolley,
2018):
• Disease Surveillance and signal detection: Precision signal detection or
disease surveillance using big data has shown efficacy in air pollution,
antibiotic resistance, cholera, dengue, drowning, drug safety,
electromagnetic field exposure, Influenza A (H1N1), Lyme disease,
monitoring food intake, and whooping cough. Disease surveillance includes
the tracking of individuals i.e., human carriers, patients, or victims through
the human movement to understand how disease transmission occurs. This
sort of surveillance results in a large volume from the stream of real-time
data generated by humans. As a result of human fast movement and the
pg. 77
spread of diseases, the real time data analysis offered by big data makes
the tracking of diseases possible.
• Predicting risk: The prediction of public health risk improves the modes of
implementing preventative measures. This prediction can be accomplished
by using big data models. An example of this prediction is Google Flu trends,
which provided info regarding flu trends in different countries. However,
Google later stopped this service.
• Target interventions: Target interventions have been applied in the
treatment of small populations within a larger population who suffer from a
certain disease. An example of this includes the treatment of gonorrhoea in
the 1980s, HIV in the 1990s, breast cancer in the 2000s, and malaria in the
2010s. However, Big data is now being used for targeted intervention that
is specific to an individual rather than the public in general. Examples of this
finer-grain treatment intervention include interventions in childhood asthma,
childhood obesity, diarrhea, Hepatitis C, HIV, injectable drug use, malaria,
opioid medication misuse, use of smokeless tobacco, and the Zika virus.
Nowadays, this extends to vaccines, where it is no longer ‘one size fits all’
but rather personalized treatment for individual outcomes.
• Understanding Disease: Data volume and variety in epidemiology have
grown consistently over time well before the age of big data. Contemporary
exponential increases in data sizes, and perhaps more importantly
increases in variety of data sources, make big data a valuable addition to
the epidemiologist’s toolkit.
Big data can play a critical role in achieving economic and efficiency targets in
healthcare (Murdoch & Detsky, 2013). There are 4 ways big data can help
achieve this in healthcare (Murdoch & Detsky, 2013):
1. Big data may greatly expand the capacity to generate new knowledge as
the cost of answering many clinical questions prospectively, and even
pg. 78
retrospectively, by collecting structured data is prohibitive. Analyzing the
unstructured data contained within EHRs using computational techniques
(eg, natural language processing to extract medical concepts from free-text
documents) permits finer data acquisition in an automated fashion.
2. Big data may help with knowledge dissemination. Most physicians struggle
to stay current with the latest evidence guiding clinical practice. The
digitization of medical literature has greatly improved access; however, the
sheer number of studies makes knowledge translation difficult. Even if a
clinician had access to all the relevant evidence and guidelines, sorting
through that information to develop a reasonable treatment approach for
patients with multiple chronic illnesses is exceedingly complex. This
problem could be addressed by analyzing existing EHRs to produce a
dashboard that guides clinical decisions. For example, this approach is
being used in the collaboration between IBM's Watson supercomputer and
Memorial Sloan-Kettering Cancer Center to help diagnose and propose
treatment options for patients with cancer. The big data approach differs
from traditional decision support tools in that suggestions are drawn from
real-time patient data analysis, rather than solely using rule-based decision
trees. For example, longitudinal diagnostic data have been shown to predict
a patient's risk of a future diagnosis of domestic abuse. Data-driven clinical
decision support tools could also lead to cost savings and help with
appropriate standardization of care. Just as purchasers receive messages
from Amazon (eg, customers like you also bought this book), clinicians may
receive messages that inform them about the diagnostic and therapeutic
choices made by respected colleagues facing similar patient profiles.
3. The capabilities of big data may assist in translating personalized medicine
initiatives into clinical practice by offering the opportunity to use analytical
capabilities that can integrate systems biology (eg, genomics) with EHR
data. The Electronic Medical Records and Genomics Network does so by
using natural language processing to phenotype patients, in an effort to
streamline genomics research.
pg. 79
4. The transformation of healthcare information by big data can allow the
delivery of information directly to the patient, which can empower them to
play a more active role. The model that’s currently in place stores patients'
records with healthcare professionals, putting the patient in a passive
position. However, this may change in the future, where medical records
may reside with patients. Big data offers the chance to improve the medical
record by linking traditional health-related data (eg, medication list and
family history) to other personal data found on other sites (eg, income,
education, neighbourhood, military service, diet habits, exercise regimens,
and forms of entertainment), all of which can be accessed without having to
interview the patient with an exhaustive list of questions. By doing so, big
data offers a chance to integrate the traditional medical model with the social
determinants of health in a patient-directed fashion. Public health initiatives
to reduce smoking and obesity could perhaps be delivered more efficiently
in this way by targeting their messages to the most appropriate people
based on their social media profiles.
Using data efficiently has the potential to improve the care of patients and lower
costs (Jones, Pulk, Gionfriddo, Evans, & Parry, 2018). The use of Big data does
not only provide more solutions but can play an important role in decreasing
healthcare costs. A 2013 report by McKinsey & Company estimated that using
big data could reduce healthcare spending by $300-450 billion annually, or 12-
17% of the $2.6 trillion baseline in annual U.S. healthcare spending (Marconi &
Lehmann, 2014).
In healthcare, big data can provide a number of benefits and initiatives that can
have a positive influence on the delivery of care. Table 2.10.2 highlights the
different lessons from Big Data in healthcare and how it is impacting healthcare.
Table 2.10.2: Lessons Learned in Applying Data to Drive Care. Adapted from (Jones, Pulk, Gionfriddo, Evans, & Parry, 2018).
Lesson Explanation
Data can inform care Continual synthesis of data can
influence treatment best practices and
pg. 80
help to individualize care and treatment
approaches.
Stakeholders should be engaged early Successful implementation of
organizational initiatives and clinical
best practices occurs when
stakeholders across the organization
are engaged in identifying opportunities
for improvement from the beginning, at
the point of decision-making, and at
final best practice.
Collaboration is necessary for complete
data
Organizations may not have access to
all the data they need to improve care
processes; thus, utilization of external
data may be necessary
but is not always complete.
Communications with stakeholders
must be open and continuous
Open and constant dialogue between all
stakeholders is necessary to ensure
uniform understanding and utilization of
data.
Data can create efficiencies Processes can be improved by creating
discrete data elements for previously
nondiscrete data fields that allow data to
be accurately and
constantly retrieved and quantified.
With big data becoming more apparent in healthcare, it is producing in new
insights and changing some of the ways care is delivered. Data used in
healthcare must display the right properties and be accurate to prevent medical
errors.
2.11 Data Accuracy
Data itself, is defined as information in the form of facts or figures obtained from
experiments or surveys, used as a basis for making calculations or drawing
conclusions as defined by (Dumas, 2013). Nowadays, data in the Information
Technology (IT) domain is often referred to as facts stored and shared
pg. 81
electronically in databases or computer applications (Sebastian-Coleman,
2013). Data in healthcare is viewed as the lifeblood of a continuous learning
health system, and the flow of such data is necessary to coordinate and monitor
patient care, analyze and improve systems of care, conduct research to
develop new products and approaches, assess the effectiveness of medical
interventions, and advance population health (Sanders, Powers, Grossmann,
& Institute of, 2013).
The significant thing about data in modern society is that it is a representation
of people, events or concepts in the real world (Sebastian-Coleman, 2013). The
representation allows for an event to be captured in digital format and stored in
databases for later representation. In the case of mHealth, the collected data is
simply patients describing their symptoms, side effects and providing other
information when suffering from an illness (Estrin & Sim, 2010).
In a health context, the quality of the data helps evaluate health, assess
effectiveness of interventions, monitor trends, inform health policy and set
priorities (van Velthoven, Car, Zhang, & Marušić, 2013). Accuracy is an element
of data quality that is intended to achieve desirable objectives using legitimate
means, and its accuracy is the original source of the data (WHO, 2014). The
international Organization for Standardization (ISO) ISO 8000 section 130
states that accuracy is a measure of how closely data represent the data of the
real world at any given time (Talburt & Zhou, 2015). In healthcare, the traditional
and standard method of collecting data from patients is through the direct
observation, also known as Gold Standard, which is a direct observation of
patients at the clinic (Eisele et al., 2013).
The data consists of multiple elements, referred to as Data Quality. Table 2.11.1
presents the different elements of Data Quality according to two definitions
collected from WHO (2014) and Batini & Scannapieco (2016).
pg. 82
Table 2.11.1: Definitions of Data Quality Elements. Adapted from ((WHO, 2014) and (Batini & Scannapieco (2016)).
Data Quality Element WHO Definition Batini and Scannapieco
(2016) Definition
Accuracy and Validity The original data must be
accurate in order to be
useful.
Accuracy is the closeness
between a value v and a
value v’, considered as
the correct representation
of the real-life
phenomenon that v aims
to represent.
Reliability Data should yield the same
results on repeated
collection, processing,
storing and displaying of
information.
Whether data can be
counted on to convey the
right information, it can be
viewed as correctness of
data.
Completeness All required data should be
present and the
medical/health record
should contain all pertinent
documents with complete
and appropriate
documentation.
The extent to which data
are of sufficient breadth,
depth, and scope for the
test at hand.
Legibility All data whether written,
transcribed and/or printed
should be readable.
N/A
Currency and
Timeliness
Information, especially,
clinical information, should
be documented as an event
occurs, treatment is
performed, or results noted.
Currency: how promptly
data are updated.
Timeliness: how current
data are for the task at
hand.
Volatility: frequency with
which data vary in time.
Consistency N/A Consistency, cohesion,
and coherence refer to
the capability of the
pg. 83
information to comply
without contradictions to
all properties of the reality
of interest, as specified in
terms of integrity
constraints, data edits,
business rules, and other
formalisms.
Accessibility All necessary data are
available when needed for
patient care and for all other
official purposes.
Accessibility measures
the ability of the user to
access the data from his
or her own culture,
physical status/functions,
and technologies
available.
Meaning or
usefulness
Information is pertinent and
useful.
N/A
Confidentiality and
Security
Both important to the patient
and in legal matters.
N/A
The Elements of Data Quality have been further elaborated on and put in a
context to understand how each element works. This has been done by Wang
& Strong (1996) who classified the elements of Data Quality in a framework
(See Table 2.11.2):
Table2.11.2: Data Quality Dimension Framework. Adapted from (Talburt, 2010).
Intrinsic Contextual Representational Accessibility
Accuracy Value-added Interpretability Access
Objectivity Relevancy Ease of
understanding
Security
Believability Timeliness Representational
Consistency
Reputation Completeness Conciseness of
Representation
Amount of Data Manipulability
pg. 84
When data lacks accuracy, currency or certainty, it can have catastrophic
results (Sadiq, 2013). For an mHealth solution to be effective and safe, the data
collected from mHealth devices, wearables and applications must be accurate
and secure (Mottl, 2014). Data accuracy is a foundational feature or dimension
that contributes to data quality (Gregori & Berchialla, 2012) and it is still a major
part of mobile health developments, as some of the traditional ways of
assessing the patients provide inaccurate data (Lin, 2013). In a study for using
Multimedia Messaging Service for the emergency teleconsultation of
orthopedia, it was found that the service demonstrated good reliability but
provided poor diagnostic accuracy (Chandhanayingyong, Tangtrakulwanich, &
Kiriratnikom, 2007). The elements of data quality can lead to serious problems
that affect patient safety. Magrabi, Ong, and Coiera (2016) have demonstrated
(See Table 2.11.3) how data can harm patients and risk their safety.
Table 2.11.3: HIT errors can lead to adverse events and patient harm. Adapted from (Magrabi et al., 2016).
Type of Error Example of error occurring in healthcare
Data are wrong Example: Mismatch in the order of medicine
Data are missing Example: Patient record missing critical information about
allergies that might impact patients during treatment
Data are partial Example: Lack of information during patient discharge,
which could be as simple as missing the dosage while
prescribing medicine
Data are delayed Example: Delay in administering medicine for patients when
instructed by a hospital physician using HIT system
Data inaccuracy can occur as a result of many factors that affect the quality of
the data. There are 4 different sources of data inaccuracy and they include (See
Table 2.11.4) (Olson, 2003):
Table 2.11.4: Source of Inaccurate Data. Adapted from (Olson, 2003).
Source of Inaccurate Data
1 Initial Data Entry Mistakes, Data Entry Processes, Deliberate, System
Errors
2 Data Decay Accuracy of data when originally created over time
3 Moving and
Restructuring
Extract, Cleansing, Transformation, Loading,
Integration
pg. 85
4 Usage Faulty Reporting, Lack of Understanding
In addition to the 4 sources listed above, intentional and unintentional wrong
data entry and speed of data collection that are misleading, can result in
misallocating resources or interventions when needed for the patients (Patnaik,
Brunskill, & Thies, 2009). Inaccurate readings, insufficient amount of data,
movement and physical activities also contribute to inaccurate data provided
through the mHealth devices (Mena et al., 2013). Another factor that affects the
quality of data is security breaches, where unauthorized modification or
alteration is made to patients’ data that compromise their confidentiality and
privacy (Mena et al., 2013). Concerns associated with data accuracy and
validity are persistent and can become a risk to patients’ safety (Linda, 2012).
For data to be of value, it must always consist of completeness, consistency,
currency, relevance and accuracy (Närman et al., 2011). In mHealth, these
elements of data quality can be compromised as data goes through 5 different
stages of collection, transmission, analysis, storage and presentation (Klonoff,
2013) before being used.
There are some measures implemented in mHealth technologies that assure
data accuracy. The WE-CARE system, is a mHealth solution for the detection
of cardiovascular diseases, in which 7 Electrocardiography (ECG) leads are
used to guarantee that data is accurate (Anpeng et al., 2014). Similarly, in a
direct observation, when the medical professional performs a certain diagnosis
such as blood pressure reading (collecting data), there’s usually other
observations made to help evaluate the patients. In a mHealth solution, a single
reading of blood pressure does not convey the full picture and is often
measured against thresholds (high and low limits) which trigger an alarm. Lack
of data does not only present a difficulty in interpreting the results but also in
making correlations and understanding deviations from certain activities
(Tollmar, Bentley, & Viedma, 2012).
The accuracy of the data contributes to data integrity that has been defined as
maintaining and assuring the accuracy and consistency of data over its entire
life-cycle (Cucoranu et al., 2013).
pg. 86
2.12 Information Integrity
In the modern world, information is power, and organizations with the most
accurate and readily available data make the fastest decisions with the least
negative impact as explained by Tupper (2011). Information at the very basic
level is raw data that is processed and transformed into information from which
knowledge is then extracted (Dumas, 2013). The integrity element of
information is about having the right properties of information including
sensitivity in which information is used, as well as encompassing accuracy,
consistency and reliability of the information content, process and system
(Fadlalla & Wickramasinghe, 2004). A similar definition has been provided by
Sinnett (2008) in which he states that information integrity has two definitions
and are as follows:
1. The dependability and trustworthiness of information
2. The accuracy, consistency and reliability of information content, process
and system.
Information integrity may be viewed as one of three main components of
“Information Value” that provides benefits specific to a consumer of the
requested information (Sinnett, 2008). The three components of Information
Value are (Sinnett, 2008):
• Usefulness, or suitability to purpose
• Usability, or its form and accessibility; and
• Integrity, or accuracy, consistency and reliability.
Integrity in a healthcare context means that an organization has the ability to
prove that the information is authentic, timely, accurate, and complete and it is
the fundamental expectation from patients, providers and other stakeholders
such as regulatory agencies (Datskovsky, Hedges, Empel, & Washington,
2015). As mHealth can be used in a number of ways, it is vital that the
information generated is accurate in order to avoid misdiagnosis, delayed care
seeking, incorrect self-treatment, conflict over appropriate care or non-
adherence to treatment plans and medication (Kahn, Yang, & Kahn, 2010).
Claude Shannon developed a theory around Information where he
pg. 87
characterized information as a message transmitted from a sender to a receiver
in a way such that the message is understood by the receiver and that the
importance of this process is the integrity of the transmission (Talburt, 2010). In
his model, Shannon differentiated between data and information, where data
are assertions about the state of a system that become information when they
are placed into a relationship to each other and interpreted by the receiver
(Talburt, 2010).
Self-management can prove critical in reducing healthcare costs while
delivering optimal healthcare services (Monsen, Handler, Le, & Riemer, 2014)
by delivering the right information to patients that do not only encourage self-
management but also empower and allow them to understand their medical
conditions (Hayes & Aspray, 2010). The effect of poor and insufficient
information can disengage people from using mHealth, causing inconsistent
and incomplete data. For instance, physical inactivity according to statistics
from WHO (2015) is the fourth leading cause of death accounting for 3.2 million
deaths globally. Wearable devices such as fitness trackers, is an example of
how doctors may promote and encourage better physical activity, yet they do
not sufficiently satisfy the patient with the information provided and therefore
can deter people from exercising (Product Design&Development, 2015).
The integrity of information produced as a result of shift in the dynamics of
technology has been getting more focus as the interaction experience has
changed (Cunningham, 2012). The shift from clinician care towards patient
centred model is encouraging patients to actively self-manage and make
decisions concerning their health (David, Sara, & Erin, 2011), and to harness
the power of mobile health in order to benefit from it, the technology must be
error free and provide rich health information.
In Health Information Management (HIT), wrong data, missing data, partial
data, and delayed data, can affect the quality of the information and they result
in information having errors that lead to medical errors (Magrabi, Ong, & Coiera,
2016). Errors in healthcare system due to data loss, incorrect data entry,
pg. 88
displayed or transmitted data can lead to loss of information integrity (Bowman,
2013).
To overcome the challenge of correctly treating patients using mHealth and
ensure information integrity, data governance, information workflow
management, internal controls, confidentiality and data privacy processes must
exist (Flowerday and Solms, 2008). These processes along with Information
Technology can improve the quality of care by decreasing medical errors due
to inaccurate and untimely information (Mahmood et al., 2012). Using a
semantic tool when processing the data and transforming it into information,
can prove crucial to detecting errors when inaccurate information is generated.
Thus, stopping it from being turned into actions and preventing medical errors.
Figure 2.12.1: The Omaha System 2005 version (Adapted from The Omaha System Chart, 2018)
The Omaha System (latest 2005 edition) is a semantic tool used in a number
of domains including medical diagnosis which involves the aggregation and
analysis of clinical data and better information management (The Omaha
System, 2015). The Omaha system has three main components, each
pg. 89
designed for a specific use. These components are described in Table 2.12.1
and the flow is shown in Figure 2.12.1.
Table 2.12.1: Components of Omaha System
The Omaha System Schemes Application
Problem Classification Environmental Domain, Psychosocial
Domain, Physiological Domain, Health-
related Behaviours Domain
Intervention Scheme Teaching, Guidance and Counselling.
Treatments and Procedures. Case
Management.
Problem Rating Scale for
Outcomes
A rating scale to assess client progress
throughout the period of service.
The introduction of Omaha System in mHealth would increase the chance of
capturing accurate data as its comprehensive problem classification stage
would enforce and bring an understanding of which relevant health data need
to be captured. The intervention component would mean empowering patients
by providing information that are relevant to their condition. It can also have a
role in detecting inaccurate data by examining the information fed back to the
patient. The rating of the outcome would provide valuable feedback in designing
the care plan. The learning from these 3 components can be incorporated into
an algorithm that can detect the flow of data and flag any erroneous data.
The detection of inaccurate data and information can be tested and done using
Machine Learning Algorithms, a method where algorithms learn from data to
produce high quality information.
2.13 Machine Learning
Smarter health is an example where big data allowed the discovery of a disease
before it was apparent (Marr, 2015). The availability of such data has revived
the field of Artificial Intelligence and Machine Learning (Kamath & Choppella,
2017). Machine Learning is a branch of Artificial Intelligence (AI) and it
combines multiple fields such as Statistics, Algorithms and Computation,
pg. 90
Database and Knowledge Discovery, Pattern Recognition and Artificial
Intelligence (Kamath & Choppella, 2017).
As a result of using smartphones for health management, this has enabled
smarter use of data as we shift from curing to anticipating and preventing
diseases before they occur through real time data analysis (Kumar et al., 2013).
With Mobile health (mHealth), the technology is seen as a great opportunity and
convenient way of tackling diseases and illness from child to chronic diseases
(Curioso & Mechael, 2010). Such smart technology has been battling chronic
diseases by using mobile phones and other devices that have multiple sensors,
as well as data connectivity that provides around the clock support (Haddad et
al., 2012). This technology is fast growing and can help solve complex problems
and diagnosis of diseases (Lev-Ram, 2012). The ability to solve complex
problems and perform diagnosis, is due to the new generation of smartphones
having more advanced computing and communication (internet access, geo-
locations) capabilities compared to older versions of mobile phones, where they
performed simple tasks and had limited capabilities (Boulos, Wheeler, Tavares,
& Jones, 2011).
Varshney (2009) describes common medical errors as those found during
investigation, diagnosis, treatment, communication and office administration
errors. Eliminating errors in mHealth can be achieved by learning about the
collected data and applying analysis techniques to find new insights that help
facilitate a better understanding of how the technology works. A common
method is using machine learning algorithms. Machine learning is a knowledge
extraction method used when there is great amount of data and it is achieved
through well designed models or algorithms that can predict for new and unseen
data (Lambin et al., 2013). The machine learning algorithms learn and improve
the outcomes through experience and observations (Oquendo et al., 2012).
This technique is different to rule based validations where rules are static and
there’s no learning. Studies in fields of e-health and telemedicine included a
pre-validation of the data during time of data collection and entry (Monda,
Keipeer, & Were, 2012).
pg. 91
Table 2.13.1: Role of Machine Learning in mHealth.
Machine Learning
Algorithm
Applications Uses Source
Artificial Neural
Networks
Diabetes Diagnosing diabetes
using neural networks
on small mobile
devices
(Karan,
Bayraktar,
Gümüşkaya,
& Karlık,
2012)
Support Vector
machines, sparse
multinomial logistic
regression (SMLR),
Naïve Bayes,
Decision Trees, k-
nearest neighbor
Fall Detection Fall Classification by
Machine Learning
Using Mobile Phones
(Albert,
Kording,
Herrmann, &
Jayaraman,
2012)
Bayesian Network Heart Rate Real-Time Robust
Heart Rate Estimation
based on Bayesian
Framework and Grid
Filters
(Radoslav &
Pavel, 2012)
Classifiers Autism Use of Machine
Learning to shorten
observation-based
screening and
diagnosis of autism
(Wall,
Kosmicki,
DeLuca,
Harstad, &
Fusaro, 2012)
Table 2.13.1 illustrates the different algorithms used in mHealth developments
that accurately detect the symptoms of a disease, perform diagnosis and
monitor patients.
The concept of Machine Learning is – learning that improves with experience
in relation to some task. That is (Bell, 2014):
• Improve over task, T
• With respect to performance measures, P
• Based on experience, E
pg. 92
Machine learning algorithms can play a pivotal role in acquiring accurate data
through pre-trained algorithms that can be deployed in mHealth solutions. The
Support Vector Machine (SVM) algorithm was deployed in a blood pressure
measurement application on an android tablet that detected the patient’s arm
and ensured stability, in order to acquire accurate reading of the data performed
by the cuffs (Murthy & Kotz, 2014). In a fall detection scenario, recorded data
were used from a database which contained 95 instances of recorded falls, from
which then four types of machine learning algorithms were applied to accurately
detect a fall (Sannino, De Falco, & De Pietro, 2014).
Machine Learning techniques can close the ‘gap’ of incomplete or inaccurate
data through the reasoning capabilities that they have (Martha & Bart, 2006). A
simple method is the use of fuzzy logic algorithms. Fuzzy logic is a method that
renders precise what is imprecise and is excellent in handling ambiguous, fuzzy
data and imprecise information prevalent in medical diagnosis (Djam & Kimbi,
2011). Fuzzy logic has been used in mimicking the regulation of glucose level
of a patient (Zhu, Liu, & Xiao, 2014). This mimicking behaviour can be used in
mHealth to enforce correct compliance and intelligently detect inaccurate data
through calculations performed by the algorithms.
This method along with other machine learning algorithms such as Artificial
Neural Networks or Association Rules, can be applied at the stage of data
collection where values captured are of exact or approximate reflection of the
patient’s current condition. The role of machine learning in detecting inaccurate
data through reasoning, could prove crucial in enhancing the quality of data
collection stage of mobile health as the accuracy aspect of data is a major
challenge in itself. Removing inaccuracy and assuring high data quality, would
mean a plethora of solutions that can be developed to help manage diseases
and lower the premature rate of deaths, as well as reduce healthcare costs.
As Machine learning is becoming more apparent in industries and used in many
applications, there were multiple frameworks created, one of which Reese,
Reese, Kaluza, Kamath, and Choppella (2017) included in the process of
pg. 93
applying Machine Learning Algorithms to different problems and applications
and that is by:
1. Data and problem definition: Asking what the problem is, the importance
of the problem and the end result of applying Machine Learning to a problem.
2. Data collection: Collecting data that will be able to address the question.
3. Data pre-processing: Cleaning data where data has missing values,
smoothing noisy data, removing outliers, and resolving consistencies.
4. Data analysis and modelling with unsupervised and supervised
learning: Data analysis and the selection of supervised or unsupervised
machine learning.
5. Evaluation: Assessing the model and how it accurately predicts the new
data. Overfitting and underfitting can occur at this stage, making the model
less useful on new data.
Machine learning has been used in a number of domains and in solving different
problems. It has been used to (Lantz, 2013):
• Predict the outcomes of elections
• Identify and filter spam messages from e-mail
• Foresee criminal activity
• Automate traffic signals according to road conditions
• Produce financial estimates of storms and natural disasters
• Examine customer churn
• Create auto-piloting planes and auto-driving cars
• Identify individuals with the capacity to donate
• Target advertising to specific types of consumers.
The learning done by the machines comes from the training performed by the
algorithms. There are several ways of training the algorithms (Reese et al.,
2017):
• Supervised learning: The model is trained with annotated, labelled, data
showing corresponding correct results.
• Unsupervised learning: The data does not contain results, but the model is
expected to find relationships on its own.
pg. 94
• Semi-supervised: A small amount of labelled data is combined with a larger
amount of unlabelled data.
• Reinforcement learning: This is similar to supervised learning, but a reward
is provided for good results.
Healthcare can benefit from Machine Learning by applying it for different
problems in healthcare. Such applications of Machine Learning in healthcare
include Evidence-based medicine, epidemiological surveillance, drug events
prediction, and claim fraud detection, using machine learning types such as
supervised, unsupervised, graph models, time series, and stream learning
(Reese et al., 2017). Machine learning has been employed by big companies
like IBM, for building machines like Watson, where it can assist different
industries such as Healthcare through its artificial intelligence using the
powerful computing behind the machine (Mohammed, Khan, & Bashier, 2016).
Table 2.13.2: Machine Learning for Healthcare: On the Verge of a Major Shift in Healthcare Epidemiology. Adapted from (Wiens & Shenoy, 2018).
Application Role of Machine Learning in infectious diseases
Predicting Risk of
Nosocomial
Clostridium difficile
Infection
Healthcare-associated infections remain at a large despite
the efforts of reducing the number of such infections. This is
in part due to a lack of an effective clinical tool that can
accurately measure patient risk. Researchers have sought to
develop models for predicting patient risk of Clostridium
difficile infection (CDI). A way of making such prediction
possible is by applying ML-driven approaches that can
successfully leverage the entire contents of the EHR. Such
clinical data contain information regarding medications,
procedures, locations, healthcare staff, lab results, vital signs,
demographics, patient history, and admission details. ML
techniques learn to map these data to a value that estimates
the patient’s probability of CDI. Although more complex than
low-dimensional tools for calculating patient risk, models that
leverage the richness of the EHR can be significantly more
accurate. Such models, based on thousands of variables,
have been extended to change over the course of an
pg. 95
admission, capturing how risk factors change over time.
These time-varying models could be incorporated into an
EHR system, using streaming data to generate daily risk
estimates for each inpatient.
Predicting
Reservoirs of
Zoonotic Diseases
ML models have been used in to predicted hypperreservoir
species for zoonotic diseases, as they account for billions of
human infections and millions of deaths per year worldwide.
The prediction included parameters such as lifespan, habitat
etc, which resulted in high accuracy of in predicting such
diseases. This in return can help in direct surveillance, vector
control and research into vaccines and therapeutic treatment
for such diseases.
Predicting Clinical
Outcomes in
Ebola Virus
Disease (EVD)
Infection Using
Initial Clinical
Symptoms
During the 2013-2016 West African Ebola Virus period, ML
was used within a limited dataset in predicting the clinical
outcomes for those infected with the disease. The data was
publicly available de-identified dataset, where they accurately
predicted outcome of infection with only a few clinical
symptoms and laboratory results.
Predicting Patients
at Greatest Risk of
Developing Septic
Shock
ML was applied in predicting septic shocks, which can help
with early interventions prior to septic shocks. The prediction
occurred at a median of more than a day prior to the septic
shock occurring, providing clinicians sufficient time to
potentially prevent disease or mitigate its severity. This
prediction goes beyond the intervention and can assist in
designing clinical trials.
Table 2.13.2 is another example of how Machine Learning has been applied in
the health domain, highlighting the role of the algorithms and the solutions it is
providing healthcare professionals with. The applications of Machine Learning
are endless and with proper training, they can be applied in a number of
domains.
Chapter Summary:
Healthcare delivery is a big domain with multiple types of care for different
conditions. The domain is complex, as it has a number of issues that affect the
pg. 96
delivery of healthcare services. Technologies like IoT present multiple
opportunities and gateways for delivering health services and care using the
technology that benefit the healthcare domain and lower the pressure on the
healthcare systems. The cases of chronic diseases present a big pressure on
healthcare systems. However, mHealth can reduce the pressures and deliver
more personalized care for people with chronic diseases. This is all dependent
on data and the different elements data have. Data and Information are keys to
facilitating successful patient treatment and enabling self-management for
reducing costs and pressure on the healthcare systems, yet they are prone to
errors. Machine Learning is one of the options where technology can create
successful solutions that can learn from data and detect errors when or before
occurring, ensuring medical errors are reduced and prevented from happening.
pg. 97
Chapter 3 : Research Methodology Research design is ‘the logical sequence that connects empirical data to a
study’s initial research questions and, ultimately to its conclusions’ as explained
by Yin (2014). In addressing this study’s research question, a qualitative
research method is applied, and it involves the use of an exploratory case study.
The qualitative, exploratory case study focuses on examining empirical
secondary data from an existing mHealth solution for diabetes, with the data
being accessible for research purposes. The following describes the logical
sequence for addressing this study’s research question.
3.1 Single Case Study
Case study is ‘an empirical inquiry that investigates a contemporary
phenomenon (the “case”) in depth and within its real-world context, especially
when the boundaries between the phenomenon and context may not be clearly
evident’ (Yin, 2014). Case studies are not considered a methodology but rather
a choice of what is to be studied (Denzin and Lincoln, 2011), and they are suited
for studying a single group, event or person (Donley, 2012). Cases can be an
individual (group, family, class), an institution (school, factory) or a community
(town, industry) (Gillham, 2000). The selected case study for this research is a
mHealth solution for the treatment of people with diabetes, with the case being
patients’ data. It is a single case study and the selection of this case study is
guided by two principles. First, the research question guides what case is to be
selected and studied (Corbin and Strauss, 2008), with the research question
posed in this study being:
‘How Can Machine Learning be applied in mHealth Solutions to address Data
Accuracy and Information Integrity?’
This research question is one that conforms to Yin’s (2014) criteria for selecting
a case study, as the form of the question is ‘How’, requiring no control of
behavioural events and focuses on contemporary events. As the study focuses
on understanding how data inaccuracy can occur in mHealth and its effect on
information integrity, the case (patients’ data) will be studied in its real-world
context (using the data for the treatment of diabetic patients). The case under
pg. 98
study required no control of behaviour in order to produce the medical data,
which permits the case to be studied and explored as it had occurred rather
than controlling how it occurs.
The second principle in selecting such case study is the single-case study
rationale where the case is critical to the theory (Yin, 2014) and relevant to the
research question. mHealth for diabetes is critical to the theory proposition as
it allows for the theory to be tested. The accuracy of the data plays a major role
in the delivery of healthcare services via mHealth.
With chronic diseases projected to increase, patients’ data are critical for the
delivery of effective mHealth services for the treatment of people with a chronic
disease. Patients’ data can present a complexity in treating patients due to the
shift in delivering healthcare services using mHealth. The change from being
treated at the clinic to self-management of health using mHealth technology,
could allow a gap for errors. Accessing such case study, enables the research
question to be addressed by examining patients’ data and exploring the
characteristics of the data, the intended meaning when data was produced and
how it contributes towards the treatment of the patient.
In health-related studies, using a case study to address a research question is
due to the nature of complex and challenging environments in healthcare that
require methods that do not only explore and understand the process, but also
provide diverse and contradictory perspectives (Caronna, 2010). This
complexity is dealt with through case studies as they allow for the exploration
of processes, activities and events associated with a phenomenon (Creswell,
2014) through detailed observation of information from records (Bowling, 2009).
When studying a new emerging technology such as mHealth, in particular the
core of the technology, data, where its value goes beyond the treatment of
patients, is complex. Gaining new insights into a new phenomenon where little
or no knowledge is available can be achieved through exploratory designs
(Steinberg, 2015). Exploratory research designs begin with the research
questions (Steinberg, 2015). Literature pertaining to data quality in information
systems is extensive, including the standards that have been developed. With
pg. 99
mHealth still in its early stages, there’s little known literature about the accuracy
of the solutions being developed and the accuracy once the solution is live,
where it is out of the lab. Figure 3.1.1 shows the logical flow of the study and
how it aims to answer the research question.
Figure 3.1.1: Research Design
In exploring this case study, the data collection method is based on secondary
data accessible for research purposes.
3.2 Data Collection
Digital Data Collection allows for the research question to be addressed by
tailoring the data to the research goal (Griffiths, 2009). With the case study
defined as an mHealth solution for the treatment of people with diabetes, and
the case being patients’ data, the collected data is purposefully selected to
Literature Review
Generate Codes for Thematic
Analysis from the Literature
Review
Access deidentified
data
Code the data using the
codes from the Literature
Review
Analyze the Themes using Thematic and Hermeneutic
Analysis
Identify ML Algorithm to address the
gaps
Revise Conceptual
Model
Develop Conclusion with Thesis
pg. 100
complement the case study. Grbich (2013) states that data should come from
the real world, situations or people who are closely tied to the research question
or it is the reason the research question was framed in a particular form. Thus,
the chosen method of data collection seeks data that present a chronic disease
that is relevant to the case study, it is produced by people in real world and is
authentic. The type of data collected for this study is qualitative secondary, de-
identified data of patients with diabetes. Secondary, de-identified data is data
that is used for research purposes and do not identify or represent a person
(McGraw, 2012). The de-identified data are records of patients who have
diabetes and contain information such as time and date of the readings, glucose
reading and a description of the reading. When the research design is an
exploratory design, the collection of the data is to access information
(Steinberg, 2015). Using such data, it allows the study to enter a real world
situation where the data is used to treat a serious chronic disease and the
method used to collect the data could be erroneous. It is also to understand
how diabetes work, explore the different characteristics of the disease and data
types with their meaning in terms of treatment, examining the accuracy of the
data and understanding the effects data inaccuracy has on information Integrity.
The adequacy of the data must be according to the research framework. The
purpose of using the secondary data is to further interpret the data other than
what it was originally intended for (Singleton, 1988) and to address this study’s
research question through secondary analysis of data (Seale, et al., 2007).
Analyzing the secondary data using a new research method and approach,
establishes a new conclusion different to the one reached in the original
analysis of the data (Boslaugh, 2007).
A primary data collection method is not applicable as the aim of the study is to
understand the contemporary phenomenon, that is the accuracy of data
produced in mHealth. Accessing secondary data allows for understanding of a
contemporary phenomenon by accessing data that has been produced. That is
through natural interaction and not one that is produced under this study. This
data then enables the research question to be tested and addressed. The
secondary data is accessed using an archival data source (Yin, 2014) of
patients with diabetes from a clinical solution for their treatment. Archived data
pg. 101
are considered a rich and unique source of data that is not often used (Seale,
et al., 2007). Accessing such source provides an opportunity to test real data
sets in which it was produced for dealing with serious chronic disease such as
diabetes. The data present instances (cases) that are relevant to the research
problem.
3.3 Data Sampling
With the type of data used in this study being secondary data, the sampling
technique employed in this study is convenience sampling. Convenience
sampling is a technique where the data is readily available and accessible for
the study (Bowers et al., 2011) that can be conveniently recruited (Gideon,
2012). The convenience sample is patients with diabetes where the treatment
of patients is done through the use of technology. This type of sampling allows
for the processes under study to be accessible by seeking events where they’re
most likely to occur (Silverman, 2010), that is to access information and not
generalization (Steinberg, 2015).
The convenience sample is selected to fit the research question. The first
sample of cases will be complete and accurate datasets. This is to match the
data quality components to the dataset to test whether if all data quality
components are present in the health data. The second sample of cases will be
incomplete and inaccurate data where the values from the data are not
complete. The intention of selecting the two cases is to explore the components
of data quality and how they affect the meaning of the data if they are missing.
This will then build an understanding on how inaccuracy can occur from which
then a Machine Learning Algorithm can be modelled that best represents the
case so that it can be tested.
Convenience sampling is a non-probability-based sampling technique (Fawcett
and Garity, 2009) as the data is purposefully selected to understand the
phenomenon and it is not representative of the whole population. Purposive
sampling is achieved by selecting typical cases, that is people with an illness
pg. 102
whose data is recorded in digital format for the treatment of their disease
(Steinberg, 2014).
Purposive sampling is “to purposefully select cases that are special,
troublesome or enlightening” (Emmel, 2013, p. 37). That is to not randomly
select cases but to access cases that provide information about the research
question and the phenomenon. The datasets provide information that are
critical to the treatment of patients and to the quality of care provided through
mHealth. The purposefully selected cases provide rich information about the
lived experiences of the patients.
3.4 Data Triangulation
The accuracy of the findings and interpretation of the data is further
strengthened through the use of a triangulation method to enhance the
credibility of the study (Yin, 2016). The triangulation method used for this study
is triangulation of different data sources of information through the examination
of different sources of evidence and building a coherent justification for themes
(Creswell, 2009). The data is triangulated using two strategies. First,
examination of different patients’ datasets allowing the study to explore the
characteristics of the different datasets of patients. Second, different Machine
Learning Algorithms are tested, and their results compared in order to find the
best suitable and accurate algorithm that best detects the errors in data.
3.5 Data Analysis
Qualitative case studies search for the meaning and understanding of a
phenomenon with a descriptive end product as Merriam explains it (2014).
Accessing the secondary data does not have a meaning, as Kim (2013) states
‘Texts do not mean, but we mean with text’. Understanding the meaning of the
mHealth data and its intended use within the context of medical treatment,
means taking the data and using it to seek a diverse interpretation of the texts
using the right analysis strategy. It is important to distinguish between the
analysis strategy and analysis technique. The analysis strategy for this study is
pattern matching while the two analysis techniques are Thematic and
Hermeneutics analysis.
pg. 103
3.5.1 Thematic Analysis
Thematic analysis is the coding of data into organized segments of texts before
bringing meaning to information (Creswell, 2009), and later underlining them for
generating themes that describe passages in the data (Cohen et al., 2000). The
secondary data is analyzed using Thematic analysis whereby the data is coded
into categories that represent:
1. Element of Data Quality as defined by the World Health Organization
2. Sources of Data Inaccuracy
3. Different parts of the research question (Data Accuracy, Information
Integrity, Machine Learning and mHealth).
4. New and emerging themes
This pattern matching strategy addresses the whole of the research by
understanding how the secondary data relates to each of the research
components and the literature covered in this study. In generating themes, it
forms the basis for applying a Hermeneutics technique to provide a detailed
description of the text to capture and communicate the meaning of the lived
experience for the sample being studied (Cohen et al., 2000).
3.5.2 Hermeneutic Analysis
We move from text to meaning through interpretation. Hermeneutics is
extracting meaning from silent text, that is text which has been separated from
the original authors as the authors were not present at the time text was access
(example:The Hoy Bible) . The authors are not present and the only method to
access the meaning of their experience is through interpretation. The
interpretation process involves bracketing (Grbich, 2013) where it is removing
one’s own meaning of preconceived ideas and concepts. The extraction of
meaning from text is through the coding process, where data is coded and
placed in the right category or concept that is relevant to the phenomena.
Applying Hermeneutics when analyzing the themes is to understand the
meaning of the texts rather than the cause and the effect of the text (White and
pg. 104
McBurney, 2013). Hermeneutics is comprised of three different layers; the Text,
Translation, and Interpretation Layers (Kim, 2013). In this study, the text layer
is the collected data of patients with diabetes where the data is authentic and
credible, and one that represents an instance of the lived experience. In
Translating the texts, The Omaha System and the Data Dictionaries enable the
translation of the original data through the mapping of data using the Omaha
System treatment plans and the definitions listed in the dictionaries.
Interpretation of the texts starts when the data is translated, and that is
explaining the texts from a wide array of perspectives (World Health
Organization definition of Accuracy, Omaha System Treatment plans, Output
produced by Machine Learning).
The data is analyzed by first reviewing the secondary data of patients with
diabetes. Accessing the secondary data allows for the first step of Thematic
analysis (getting familiar with the data) and also the first step of Hermeneutics
(turning to the nature of the lived experience) to be applied. The data is then
revisited and it is aligned with codes from the literature review (Data Quality,
Information Properties, mHealth, Machine Learning). The data is then mapped
with the Client Care Plan from The Omaha System to confirm the accuracy of
the data by matching the data with the treatment plan. Upon the completion of
mapping the data to the care plan, a Machine Learning algorithm is identified
that best matches the data. In order to test the Machine Learning Algorithms,
the secondary data is divided into training and testing datasets. This then
allowed for the Machine Learning algorithms to be trained using the training
datasets. Once the training was completed, the Machine Learning Algorithms
were tested using the testing datasets and the performance of the algorithms
are both captured to compare the results and the accuracy of the algorithms.
At each stage of the data analysis, Thematic analysis was applied to capture
themes to understand the meaning of the data and accuracy of the values. This
is completed by searching for themes, reviewing and defining them. These
steps of Thematic analysis built the foundation for Hermeneutics analysis as
the themes were used to reflect on what the themes meant, how they
characterized the phenomenon and allowed for the writing of the results to be
completed by maintaining a focus on the phenomenon through writing. Writing
pg. 105
of the results as the data is analyzed produces the report (last step of Thematic
analysis) and the report included the different parts and entire research (last
step of Hermeneutics analysis) including the literature review, the research
question, the collected and the analyzed data. Figure 3.5.1 illustrates the
process for analyzing the data by starting with the Text Layer that is primarily
concentrated on the text (data) and wherever data existed or used in the
research. The Translation layer is where Thematic Analysis was used to define
and name the themes by reading the data to extract meaning of the data and
initiate the steps required for the last step of Hermeneutics analysis, the
Interpretation Layer. The interpretation layer was dependent on the themes that
were generated through Thematic analysis as the data itself did not have a
meaning. The interpretation of the themes brought out the meaning behind the
data and explained what the data was used for and what the accuracy of the
data meant in the context, treating the patients.
Chapter Summary:
The research question is based on Qualitative analysis and the case study
selects a chronic disease where mHealth technology is delivering services and
applications through the phone. Thematic and Hermeneutics analyses both aid
in gaining an understanding of the domain and the results produced by them.
pg. 106 pg. 106
Figure 3.5.1 Analysis Design
pg. 107 pg. 107
Chapter 4 : Data Analysis and Findings
This chapter shows the process of analyzing the collected data and presents
the findings. The collected data consisted of structured glucose readings data
from patients suffering from diabetes. The data were collected using an
automatic electronic recording device and paper records. The purpose of using
such dataset is to see how Machine Learning can be applied to health data to
achieve a certain level of data accuracy that can deliver information with
integrity.
4.1 Domain Description and understanding of Diabetes
Understanding the health domain and how diabetes works, is important in the
analysis of the data and in interpreting the results. The following describes the
domain and how diabetes functions.
Diabetes “is a complex chronic metabolic/endocrine disorder characterised by
loss of normal blood glucose regulation and hyperglycaemia” (Whettem, 2012,
p.1). Glucose extracted from diet becomes the body’s main source of energy
(Whettem 2012). Beta cells in the pancreas produce the hormone insulin, which
is then absorbed through the body’s tissues that use glucose for energy or
stored as glycogen in the liver/muscles (Whettem, 2012). This is demonstrated
in the secondary data as it contained a domain description of how insulin
functions, how exercise affects insulin, diet, and glucose concentration (Dua &
Karra Taniskidou, 2017). The following is an extract from the domain
description included in the datasets (Dua & Karra Taniskidou, 2017):
• INSULIN
One of insulin's principal effects is to increase the uptake of glucose in many of
the tissues (e.g. in adipose/fat tissue) and thereby control the concentration of
glucose in the blood. Patients with Insulin Dependent Diabetes Mellitus (IDDM)
administer insulin to themselves by subcutaneous injection. Insulin doses are
given one or more times a day, typically before meals and sometimes also at
bedtime. Many insulin regimens are devised to have the peak insulin action
pg. 108
coincide with the peak rise in BG during meals. In order to achieve this, a
combination of several preparations of insulin may be administered. Each
insulin formulation has its own characteristic time of onset of effect (O), time of
peak action (P) and effective duration (D). These times can be significantly
affected by many factors such as the site of injection (e.g. much more rapid
absorption in the abdomen than in the thigh) or whether the insulin is a human
insulin or an animal extract. The times I have listed below are rough
approximations and I am sure that I could find an endocrinologist with different
estimates.
Regular Insulin: O 15-45 minutes P 1-3 hours D 4-6 hours
NPH Insulin: O 1-3 hours P 4-6 hours D: 10-14 hours
Ultralente: O: 2-5 hours. P (not much of a peak) D 24-30 hours.
• EXERCISE
Exercise appears to have multiple effects on BG control. Two important effects
are: increased caloric expenditure and a possibly independent increase in the
sensitivity of tissues to insulin action. BG can fall during exercise but also quite
a few hours afterwards. For instance, strenuous exercise in the mid-afternoon
can be associated with low BG after dinner. Also, too strenuous of an exercise
with associated mild dehydration can lead to a transient increase in BG.
• DIET
Another vast subject but (suffice it to say for the purposes of users of the data
set) in brief: a larger meal will lead to a longer and possibly higher elevation of
blood glucose. The actual effect depends on a host of variables, notably the
kind of food ingested. For instance, fat causes delayed emptying of the stomach
and therefore a slower rise in BG than a starchy meal without fat. Missing a
meal or eating a meal of smaller than usual size will put the patient at risk for
low BG in the hours that follow the meal.
• GLUCOSE CONCENTRATIONS
pg. 109
BG concentration will vary even in individuals with normal pancreatic hormonal
function. A normal pre-meal BG ranges approximately 80-120 mg/dl. A normal
post-meal BG ranges 80-140 mg/dl. The target range for an individual with
diabetes mellitus is very controversial. Generally, it would be very desirable to
keep 90% of all BG measurements < 200 mg/dl and that the average BG should
be 150 mg/dl or less. Note that it takes a lot of work, attention and (painful) BG
checks to reach this target range. Conversely, an average BG > 200 (over
several years) is associated with a poor long-term outcome. That is, the risk of
vascular complications of the high BG is significantly elevated.
Hypoglycemic (low BG) symptoms fall into two classes. Between 40-80 mg/dl,
the patient feels the effect off the adrenal hormone epinephrine as the BG
regulation systems attempt to reverse the low BG. These so-called adrenergic
symptoms (headache, abdominal pain, sweating) are useful, if unpleasant,
cues to the patient that their BG is falling dangerously. Below 40 mg/dl, the
patient's brain is inadequately supplied with glucose and the symptoms become
those of poor brain function (neuroglycopenic symptoms). These include:
lethargy, weakness, disorientation, seizures and passing out.
This description provides an understanding of how diabetes works, how the
datasets are obtained and what the meaning of the actual datasets is. This
description of the domain is taken into consideration when building the Machine
Learning model and in describing the secondary data.
4.2 Describing the secondary data
The secondary data were obtained from a Machine Learning Data Repository
provided by the University of California, where data is available for further
analysis and testing (Dua & Karra Taniskidou, 2017). The data consisted of 70
patients records (datasets) and each record presented an individual patient.
The description of the data is as follows:
Diabetes patient records were obtained from two sources: an automatic
electronic recording device and paper records. The automatic device had an
pg. 110
internal clock to timestamp events, whereas the paper records only provided
"logical time" slots (breakfast, lunch, dinner, bedtime). For paper records, fixed
times were assigned to breakfast (08:00), lunch (12:00), dinner (18:00), and
bedtime (22:00). Thus, paper records have fictitious uniform recording times
whereas electronic records have more realistic time stamps.
The data Diabetes files consist of four fields per record (See Figure 4.2.1). Each
field is separated by a tab and each record is separated by a new line.
File Names and format:
(1) Date in MM-DD-YYYY format
(2) Time in XX:YY format
(3) Code
(4) Value
Figure 4.2.1: Example of the Secondary Data in its original format. Example is from Data1
The secondary data also had an explanation for the code attached to each of
the readings. The explanations are as per Table 4.2:1.
Table 4.2.1: Codes and their explanations
Code Explanation
33 Regular insulin dose
34 NPH insulin dose
35 UltraLente insulin dose
48 Unspecified blood glucose measurement
57 Unspecified blood glucose measurement
58 Pre-breakfast blood glucose measurement
59 Post-breakfast blood glucose measurement
60 Pre-lunch blood glucose measurement
pg. 111
61 Post-lunch blood glucose measurement
62 Pre-supper blood glucose measurement
63 Post-supper blood glucose measurement
64 Pre-snack blood glucose measurement
65 Hypoglycemic symptoms
66 Typical meal ingestion
67 More-than-usual meal ingestion
68 Less-than-usual meal ingestion
69 Typical exercise activity
70 More-than-usual exercise activity
71 Less-than-usual exercise activity
72 Unspecified special event
The data was chosen as:
1. It presents a scenario where diabetes data is collected both using an
electronic device and recorded manually, where the collection method is
one similar to the ones discussed in the literature review. This method of
entry makes data prone to error during the data entry stage of data
collection.
2. The data is part of Machine Learning data collection, which makes it
appropriate for this research question and the techniques used in this study
provides a fresh perspective on how Machine Learning can be applied in
modern technologies that deal with chronic diseases.
3. The data is diabetes, a chronic disease that is discussed in the literature
review and a global burden to the healthcare systems around the world.
The data is not based on a mHealth solution, but it is one that can closely
simulate mHealth solution for diabetes.
4.3 Review of secondary data of patients with diabetes
The initial analysis of the data was reviewing the data in its original format and
the aim of this is to gain information about the type of data being generated by
patients with diabetes. The elements that were present in the datasets were:
pg. 112
• Completeness: This element of the data quality was present in the data as
the element was tested during the analysis for when data showed complete
datasets and where records were missing.
• Currency and Timeliness: This element of the data quality was present in the
data as the element was tested during the analysis by checking how current
the data is and the time stamps when data was recorded. The data showed
both currency (how promptly data are updated) and timeliness (how current
data are for the task at hand).
• Legibility: This element of the data quality was present in the data as the data
was readable, except for instances where data was missing correct data
format. Legilibty means the content of the data are sufficient for the data to be
interpreted.
• Consistent: This element of the data quality was present as the data showed
both instances of consistent and inconsistent datasets.
• Reliability: This element of data quality was present in the datasets as some
of the datasets yielded the same results on repeated collection, processing,
storing and displaying of information.
• Meaning and usefulness: This element of the data quality was present in the
datasets as the Information from the datasets were both useful, or in some
instances lacked information.
• Accuracy and Validity: The data demonstrated that not all elements of data
quality were available in the datasets. The following elements of data quality
were not applicable as they could not be tested during the analysis.
• Accessibility element: This element of the data quality is of limitation in this
study as the data was accessible at the time when the data was analysed.
Therefore, accessibility can be evaluated when the researcher analyses data
pg. 113
that is observed when the patient sends/transmits the data to the medical
professional, which is then when it can be tested for accessibility.
• Confidentiality and security: This element of the data quality is of also a
limitation in this study as the data cannot be tested for confidentiality as it was
secondary data and it does not identify a person. It also could not be tested
for security as the data was available for research purposes and it was not
confidential.
While some of the elements of data quality were missing, the structure of the
data was simple and had very little information around the readings. The
structure included date of the reading, time of the reading, a code explaining
the reading and the actual reading. Figure 4.3.1 is an example of the raw data
and this example is from Data20, one of the datasets.
Figure 4.3.1: A snippet of the structure of the raw data from Data20 in its original format
The structure of the data was transformed into a different format for multiple
reasons:
1. The structure of the data was too simple when algorithms were tested
against the data as the results were simple and the data was easily
classified, which defies the purpose of the study.
2. The structure also shows the human behaviour and the management of the
diseases through a sequence of data put together in a row that shows the
researcher what is happening with the patient and what the symptoms
mean, the treatment and the readings from the patient. This process allows
the researcher to follow the history of the patient and understand how the
pg. 114
patient reacts to certain aspects of the disease and how they treat their
symptoms, bringing the research question closer to being addressed.
3. The new structure of the data allowed to introduced a new complexity in
solving the issue of accuracy with data. The new structure (Figure 4.3:3)
included more meta-data that enhanced the quality of the algorithms and
made it more meaningful than before. The structure presented a simple
checklist that checked for each element of data quality for each row of the
datasets and calculate the accuracy.
Figure 4.3.2 is the structure of the data prior to the change in the structure as
per Figure 4.3.3.
Figure 4.3.2: Example of old data structure from Data1 before it was transformed into new format
Figure 4.3.3: Example of the new data structure from Data1
The new structure of the data provides more information that assisted in
deciding the algorithms and provide a closer view of the real world of health
data, data generated by humans and not systems.
4.4 The transformation of data before Machine Learning
Part of preparing the data for machine learning, the structure of the data was
transformed into a different structure in order to address the research question
and make it ready for Machine Learning. The change in the structure of the data
pg. 115
is to include metadata that can yield more information and make sense of what
the data means, thus making it relevant to the research question.
The structure of the data included headings for each of the columns as the data
must be unique prior to using Machine Learning Algorithms. The headings
were:
1. Date
2. Time of the reading
3. Code for reeading
4. Reading
5. Completeness
6. Currency and Timeliness
7. Legibility
8. Consistency
9. Reliability
10. Meaning and Usefulness
11. Accuracy and Validity
The selection of these fields is based on the list of a priori codes generated from
the literature review. Prior to the analysis of the data, a list of a priori codes was
generated from the literature review which were used in the labelling of the data.
The selection of these codes was (See Table 4.4.1)
1. Relevant to the research question, such as elements of data quality and
aspects of information integrity
2. They would provide more information about the disease, the patients, and
the management of the disease.
Table 4.4.1: List of A priori codes generated from the literature review
A priori codes generated using Elements of
Data Quality
A priori codes generated using causes
of lack of Information Integrity
Accuracy and Validity Inaccurate Data
Reliability Incomplete Data
Completeness Incorrect Data Entry
Legibility Security
Currency and Timeliness Untimely Data
Consistency
pg. 116
Meaning or usefulness
This list of a priori codes was used in the labelling of the data, which assisted
in building a machine learning model in Weka.
4.5 Labelling the data
The data were labelled using the apiori codes from the literature review. The
codes were chosen as they represented the elements of data quality, as they
can help in identify which data can be classified as accurate or not.
The order for the labelling was done by:
1. Completeness: This field tags whether the data is complete or not.
2. Currency and Timeliness: This field tag whether the data is current and
timely or not.
3. Legibility: This field tags whether the data is eligible for use or not.
4. Consistent: This field tags whether the data is consistent or not.
5. Reliability: This field tags whether the data can be relied in for making
decisions or is of no use.
6. Meaning and Usefulness: This fields tags whether the data is meaningful
and is useful to be used in decision making.
7. Accuracy and Validity: This field tags whether the data is accurate and valid
once the data has been checked for earlier check or not.
8. OverallAccuracy: This field tags whether the data overall accuracy is
correct or not. This field is of important value to Machine Learning as it is
the target variable the algorithms try to predict.
After applying the codes to the data, the datasets were given either a ‘yes’,
which marked the data as a valid dataset containing the correct data quality
element, or a ‘no’, where data did not contain the valid data quality element.
The labelling of the data follows a logic of which data is tested against each
element of data quality (See Figure 4.5.1).
pg. 117
Figure 4.5.1: Labelling of the data
By labelling the data, it became more relevant to addressing the research
question in this study as the process of the labelling demonstrated how Machine
Learning can be applied in mHealth solutions. This also assisted in defining the
problem that was later applied in Machine Learning.
4.6 Defining the problem and understanding the data
Before analysing the data using an algorithm, a problem definition was
described that assisted in the selection of the data. Applying an algorithm is not
a simple process and it requires data to be of appropriate format and quality.
The problem was defined as ‘Ways in which Machine Learning can detect data
inaccuracy and/or improve the data quality in mHealth technology’. This moved
the focus from finding the optimum algorithm that can detect the inaccurate data
at 100% accuracy but rather focus on ways in which the technology can improve
mHealth solutions by using the data and the advances in algorithm for a better
solution.
After the problem was defined, the early stages of the analysis, the datasets
were then examined to form an understanding of the data structure, type of
data, frequency of the readings and the length of the data (Figure 4.6.1 is a
snapshot of what the data looked like). The data consisted of 70 patient records,
and fields were Date Stamp, Time Stamp, Code to explain the condition, and a
reading.
Figure 4.6.1: Data Samples from Diabetic Dataset
pg. 118
The data were then re-written in a table format that simulated the patient dealing
with diabetes on a daily basis. This improved the understanding of what the
data meant, and the actions taken by the patient in different scenarios caused
by glucose levels and activities carried by the patients. It also simulated the sort
of behaviour/interaction patients have when self-managing their disease and
health. It depicted how monitoring is done, the patient’s behaviour, and the
quality of the data being generated by the patient and the disease.
This led to the generating a machine learning model that could be applied to
the datasets.
4.7 Building the Model and Selection of Weka Machine
Learning Tool
With the data labelled with values and the target value, a machine learning
model was built based on the new structure of the data (See Figure 4.4:3). The
logic of the model is one that checks for the value of each of the elements of
data quality and uses the target variable to predict whether the data is accurate
or inaccurate. The model is illustrated in Figure 4.7.1 in which it
1. Checks if data is complete or incomplete (Yes, No): Some of the data were
complete and some incomplete, thus assigning either a ‘Yes’ or a ‘No’ to
data.
2. Checks if data is current and Timely (Yes, No): All the data were current and
had a timestamp, thus assigning a ‘Yes’ to all records.
3. Checks if the data is Legible (Yes, No): All the data were in readable format,
thus assigning a ‘Yes’ to all records.
4. Checks if the data is consistent or inconsistent (Yes, No): Some of the data
were consistent and some inconsistent, thus assigning either a ‘Yes’ or a
‘No’ to data.
5. Checks if the data can be relied on for decision making (Yes, No): Some of
the data were reliable and some weren’t, thus assigning either a ‘Yes’ or a
‘No’ to data.
6. Checks if the data is meaningful and useful (Yes, No). All the data had
meaning and were useful, thus assigning a ‘Yes’ to all records.
pg. 119
7. Checks if the data is accurate and valid (Yes, No): Each instance of the data
were accurate for the specific use and valid as well. The data were given a
‘Yes’. However, the Overall Accuracy, is the last variable that took into
consideration all the elements of data quality and produced either a ‘Yes’ or
a ‘No’.
Figure 4.7.1: Machine Learning Model
Figure 4.7:2 illustrates the process in building the machine learning model. This
model was built based on data accuracy, where accuracy takes into
consideration the different elements of data quality in a decision.
Figure 4.7.2: Process of building a Machine Learning Model 1.2
Figure 4.7.2 and Figure 4.7.3 illustrate the behind the scene process for building
the actual model in Weka Machine Learning Tool and the thinking behind it.
pg. 120
Figure 4.7.3: Process of building the model 1.2
The Machine Learning model was built using Weka, an open source Machine
Learning tool provided by the University of Waikato, New Zealand. It is an easy
to use, yet powerful tool, that provides a graphical user interface for performing
machine learning. It also contains the necessary tools for performing the ‘Pre-
processing’ of data prior to running the algorithm against the data.
The algorithm used in this study is the J48 Classification Algorithm, a decision
classification machine learning algorithm that produces an analysis of the input
data and produce an output, modelling the output based on a decision tree.
This algorithm is a common one used in many technologies, including
commercial tools like music and entertainment industry.
Once the model was built, the data was then ready for the ‘Pre-processing’
stage of Machine Learning algorithm.
4.8 Data Pre-processing
Prior to analysing the data using an algorithm, the data went through a pre-
processing stage, where the data were examined, organized in a sequence
•Understand the
diabetes data
•Review the fields of
the datasets
•Tag the algorithms
Specification
•Model the data based
on elements of data
quality
Modelling•Use J48 Algorithm to
classify the data using
the target variable ‘Overall Accuracy’
Classification
pg. 121
based on daily management of diabetes, removed missing and redundant data
that might result in a biased output using the algorithms. The data was then
split into a training and testing dataset following a 70:30 rule, where 70% of the
data is used for training and 30% for testing (Zinoviev, 2016). This rule of
splitting was also applied in a similar study for classifying diabetes data where
70% of the data was used for training and the other 30% of the data for training
(Nabi, Wahid, & Kumar, 2017). The data was split using the built in
unsupervised filter (RemovePercentage) where 30% of the data was removed
which produced the 70% of the training data. The rule was later removed after
the training datasets were generated and invertSelection was set to true which
generated the 30% of testing datasets.
The pre-processing of the data ensured the data was ready for analysis and a
list of outputs was produced through the algorithms.
pg. 122 pg. 122
4.9 Findings
The following presents the findings from the Thematic analysis of the data and the outputs from the Machine Learning Algorithms.
The Thematic analysis was conducted using Nvivo, where data was coded against the codes from the literature review (a priori
codes). Thematic analysis also resulted in new themes emerging form the analysis that provided a new insight into what the data
meant.
The results from the Machine Learning training and testing were recorded based on the total number of records per dataset (patient
records), and the 70% of data used for training and the 30% used for testing.
4.9.1 Thematic Analysis
Theme Area of mHealth Findings
Expected Theme Data Quality Inconsistency
• Some of the data were inconsistent on hourly basis (when data was recorded/collected)
• The readings were sent at irregular intervals where some followed a consistent manner
• Some of the readings were inconsistent in the number of readings that were sent (some instances of when data
was recorded, it included 1 instance of a reading, other times 2 or more readings
• Inconsistency between patient readings
• Inconsistency between the readings for a patient on hourly and daily basis
pg. 123
Currency
• All the data were current as the timestamps provided evidence that the data were current for the purpose the
data were collected for
• This element of data was consistent across the 70 datasets
Incomplete
• Some of the data included a code with no explanation as the code was not included in the code explanation
• Some of the data were incomplete as there was a single reading, where most of the instances included multiple
readings, showing the patients’ behaviour and the intended use of the data.
Readability
• Majority of the the 70 datasets were readable and the characters (data) were easily readable. This means that
the data did not contain any invalid characters (ASCII characters or the format were not recongizable).
• There were instances were data contained invalid codes which were not part of the codes supplied in the
metadata
Consistency
• Consistency was a common theme amongst the data, as some of the readings were collected in a consistent
manner, including timestamps, readings, and codes on a specific day.
pg. 124
Complete
• There were instances of data were the collected data were complete and contained large amount of information
that demonstrated the purpose of the readings and patients’ behaviour.
Incorrect data
• Instances of incorrect data were present in the data. An example is data64 where the data was repeated, it did
not exactly reflect what was happening, and the data was duplicated.
Timeliness
• Frequency of the data: The data was transmitted hourly on a daily basis.
• The readings were collected in minutes, that made them an hourly reading. .
• The hourly readings were timely, which contributed to being daily readings.
Accuracy
There were different levels of accuracy that prevailed during the analysis, and accuracy is defined based on the
context of the data. For instance, the accuracy of data meant that:
1. Accuracy of the data at the hourly interval
2. Accuracy of the data at the treatment level
3. Accuracy of the data at the daily interval
4. Accuracy of the treatment of the disease on the daily basis
Emergent Theme Readings • The frequency of the data was on hourly and daily basis
pg. 125
• The readings varied at the times they were sent and the amount of data supplied was different too. The variability
in the readings of the data makes it difficult for the algorithms to operate efficiently and the data had to be manually
examined and reviewed.
Lack of Information • Some data were missing information such as codes
• Other data were a single data collected which lacked information about what was happening and why a single
collection of data.
Gap between readings • The data at times showed a gap between readings as at times data was collected at early morning (around
9:00AM) and later collected again at 18:00PM
Human Behaviour • The human behaviour demonstrated disease symptoms and the associated behaviour users go through when
dealing with a symptom of diabetes
• The collected data displayed human behaviour where a variability in the readings of data showed that the data
was transmitted at different frequencies based on the activity the patients were undertaking.
• Each record demonstrated that individuals manage their diseases differently and the decision for insulin dosage
is not sufficiently demonstrated or elaborated on
Duplicate Records • Some of the data contained duplicate records which were of no use as the data did not contain useful information.
4.9.2 Machine Learning Outcomes
pg. 126
Dataset Total number of
records per
dataset
Number of
records for
training
Number of
accurate
classified
data
Number of
inaccurate
classified data
Number of
records for
testing
Number of
accurate
classified data
Number of inaccurate
classified data
Data1 504 353 353 0 151 151 0
Data2 380 266 264 2 114 114 0
Data3 145 101 101 0 44 44 0
Data4 151 106 106 0 45 45 0
Data5 152 106 106 0 46 46 0
Data6 65 45 45 0 20 19 1
Data7 119 83 83 0 36 36 0
Data8 133 93 93 0 40 40 0
Data9 108 76 76 0 32 32 0
Data10 107 75 75 0 32 32 0
Data11 72 50 50 0 22 20 2
Data12 91 64 62 2 27 27 0
Data13 101 71 71 0 30 30 0
Data14 58 41 40 1 17 17 0
Data15 97 68 68 0 29 27 2
Data16 97 68 68 0 29 29 0
Data17 120 84 84 0 36 36 0
pg. 127
Data18 147 103 103 0 31 31 0
Data19 140 98 98 0 42 42 0
Data20 453 317 317 0 136 136 0
Data21 245 171 171 0 74 74 0
Data22 107 75 75 0 32 32 0
Data23 106 74 74 0 32 30 2
Data24 119 83 83 0 36 36 0
Data25 91 64 64 0 27 26 1
Data26 306 214 214 0 92 92 0
Data27 463 324 324 0 139 139 0
Data28 477 334 334 0 139 139 0
Data29 617 432 432 0 185 185 0
Data30 591 414 414 0 177 177 0
Data31 336 235 235 0 101 101 0
Data32 58 41 41 0 17 17 0
Data33 126 88 88 0 38 38 0
Data34 131 92 92 0 39 39 0
Data35 131 92 92 0 39 39 0
Data36 132 92 92 0 40 40 0
Data37 152 106 106 0 46 46 0
Data38 134 94 94 0 28 28 0
pg. 128
Data39 137 96 96 0 41 41 0
Data40 Data could not be tested as the records were duplicate
Data41 251 176 176 0 75 74 1
Data42 261 183 183 0 78 76 2
Data43 294 206 206 0 88 88 0
Data44 144 101 101 0 43 43 0
Data45 137 96 96 0 41 41 0
Data46 142 99 99 0 43 43 0
Data47 153 107 107 0 32 32 0
Data48 157 110 110 0 47 47 0
Data49 191 134 134 0 57 57 0
Data50 249 174 174 0 75 75 0
Data51 152 106 106 0 46 46 0
Data52 173 121 121 0 52 52 0
Data53 128 90 90 0 38 38 0
Data54 407 285 285 0 122 122 0
Data55 607 425 425 0 182 182 0
Data56 510 357 357 0 153 153 0
Data57 104 73 73 0 31 29 2
Data58 126 88 88 0 38 38 0
Data59 126 88 88 0 38 38 0
pg. 129
Data60 128 90 90 0 38 38 0
Data61 124 87 87 0 37 37 0
Data62 140 98 98 0 42 42 0
Data63 156 109 109 0 47 47 0
Data64 Data could not be tested as the records consisted of single readings and the format of the data was incorrect
Data65 565 395 395 0 170 170 0
Data66 Data could not be tested as the records were duplicate
Data67 496 347 347 0 149 149 0
Data68 528 370 370 0 158 158 0
Data69 Data could not be tested as the records consisted of single readings
Data70 151 106 106 0 45 45 0
pg. 130
Chapter 5 : Discussion This chapter discusses the findings from the study and some key points that
relate to the research question, including emerging areas that relate to mHealth.
The main goal of this study was finding a way in which machine learning could
be applied in mHealth solutions that can detect inaccurate data and to deliver
information with integrity. This was accomplished using the J48 Algorithm after
the labelling the datasets, building, training, and testing the model, where data
were classified as either accurate or inaccurate. The results from this study
affect several areas that are integral components of mHealth and ones that
must be taken into consideration in order to understand the interpretation of the
research. These include:
• The different levels of Data Accuracy
• Understanding the human behaviour
• The role of metadata in providing more information about the data
• Contextual data and how it is providing background information about the
patient
• Using contextual data in building a machine learning model to detect
inaccurate data
• The privacy of the users and the security of the data when building an
algorithm.
5.1 The different levels of Data Accuracy
Accuracy is a term that is context dependent and the term was used to discuss
the accuracy of the diabetes data in the study. The accuracy in this study was
defined on 2 levels:
1. Accuracy of the elements of data quality, that took into consideration the
accuracy of each element of data quality and how they contribute to the
overall accuracy of the full dataset. This was then applied in building the
Machine Learning Algorithm
2. Accuracy of the information provided through the data, including the context
(patients with diabetes), the human behaviour provided through the codes
in the data and the purpose of the collected data.
pg. 131
Defining these 2 levels of data accuracy, provided an insight of where in the
solutions data can be inaccurate and what signs or symptoms hint at
problematic data. This then guided the study into understanding how data
inaccuracy occurred, what inaccurate data readings looked like, and how they
could be used as a benchmark for detecting accurate data. The inaccurate data
provided an insight of how patient safety can be compromised, what triggers
should be in place to flag when data is inaccurate and how it should be
prevented from progressing into the next stage of becoming information. In
terms of chronic diseases, the elements of data quality provided a check for
each reading that came through and flagged whether the data met the
necessary criteria to be qualified as accurate data. Once the elements of data
quality were checked for each reading, the data then progressed to the next
stage of being used as information. The integrity of the information was also
impacted by the quality of the data as some instances of the data were
inaccurate, which meant that the integrity of the information could not be
accomplished and that it could lead to serious harm. Some of the data lacked
consistency, accuracy and reliability, which are flags for information lacking
integrity.
With data and information integrity impacted by inaccurate data, the reasons
and causes of inaccurate data can be better understood by understanding the
human behaviour.
5.2 Understanding the human behaviour
During the analysis of the data, a theme that emerged was human behaviour
as the data showed that the way humans manage their diabetes varies from
one patient to another. Humans behave differently, and an interesting
behaviour is a ‘lying patient’, a patient who does no adhere to treatment
(Bardosh et al., 2017). Non-adherence is an example of how data can lack
accuracy as data is missing the truth (complete data with context that shows
what is happening). The changes in human behaviour can potentially impact
the quality of the data as data is not consistent. This type of behaviour can be
pg. 132
tracked using Machine Learning where models can be built as new situations
emerge. The tracking of human behaviour using Machine Learning might prove
critical in detecting abnormalities in data that can enhance the quality of care
through mHealth. The tracking of human behaviour can have both a positive
and a negative impact on the algorithms. On a positive side, it means that once
the human behaviour has been modelled, the algorithm can be used across
multiple datasets to predict any data collected for a specific reason. On a
negative side, the changes in human behaviour, mean that algorithms must be
refined continuously as new data comes through, which results in overfitting.
The modelling of human behaviour is not a new concept and it has been done
in separate studies for different purposes. Human behaviour has been studied
over a long time with such research including human movement in time and
space that has spanned over at least over five decades (Weiner, 1999). The
purpose of the study was to forecast future travel demand to better guide the
investment of mega-scale transportation projects, where the prediction of the
human behaviour would allow an understanding of how people travel (Weiner,
1999).
The concept of modelling human behaviour using a machine learning algorithm
for a specific purpose or solution can be effective. The modelling of the human
behaviour in this study meant that the algorithms can be applied in mHealth
solutions to help continue the development of mHealth solutions and ensuring
the quality of the data and information. With the patient-clinician interaction
experience changing within the context of mHealth, the modelling of the human
behaviour might become more common in delivering an effective solution to
patients. The precision of modelling the human behaviour can be affected by
the amount of metadata available that can add a greater meaning to the data.
5.3 The role of metadata
Metadata plays an important role in assisting people to understand data and
the meaning of the data. The metadata accompanied with the datasets provided
information that helped in establishing a Machine Learning model which created
the classification algorithm. Metadata can provide more information around the
pg. 133
context of the data. In this study, the data contained 4 columns, data, time,
reading, and the code for the data. These columns provided a context around
what time of the day the reading came through, what instigated the treatment,
the before and after glucose readings. However, if the date field contained the
day of the week, it could provide more data as to the patients’ behaviour during
the weekdays and the weekends, allowing for more information to be collected
and used in the analysis of the data. The more data we have the better the
prediction is. Metadata can be revealing as they capture large amount of data
about individuals and their behaviour (Pomerantz, 2015).
Metadata is becoming more common in modern technology such with likes of
social media where metadata can reveal information about individuals and their
social networks, connection between places, and organization (Pomerantz,
2015). Metadata is becoming an inevitable section of technology to avoid as
modern technologies keep adopting the technology. The collection of metadata
can give a greater detail about the landscape of the data and the solution, thus
building contextual data that mean something, specifically when the data is
secondary data.
5.4 Contextual data and the inclusion of other metadata
The more data there is available about what is being studied, the more it can
be understood about the content and the context of the data. This is where the
more metadata we have about the data, the better we can train the algorithms
to classify and predict results. In this research, metadata were generated apart
from what was originally provided with the data. The metadata generated
through the analysis included an insight into the life of a patient having diabetes
on daily basis, and a mapping of the Omaha System Care Plan, which provided
further insights of when a reading becomes problematic and when it can be
safely acknowledged that if the condition is within the right limit. Both of these
metadata information provided a better insight into how the data works and how
to better train the data and push towards accurate results. Metadata can help
build greater solutions using artificial intelligence such as the use of Machine
Learning to better facilitate health IT solutions.
pg. 134
5.5 Machine Learning and Artificial Intelligence
A separate perspective derived from the Artificial Intelligence world, and one
that is related to this research is Machine Intelligence. The machines can be
trained to do certain tasks and accomplish goals with rewards. Self-driving cars
need to integrate pattern recognition, with other capabilities, including
reasoning, planning and memory (Loukides & Lorica, 2016). Loukides & Lorica,
(2016). These different capabilities are to enable self-driving cars to detect
hazards, reason to understand regulations, planning to plan routes, and
memory to repeat and learn (Loukides & Lorica, 2016). In healthcare, it is a
similar case to the AI solution as the algorithms must cater and be adequate for
different diseases and patients. Figure 5.5.1 presented below is an illustration
of this process and shows how algorithms can be applied to different areas of
data in mHealth to achieve a higher level of data accuracy and information
integrity.
Figure 5.5.1: Application of Machine Learning in mHealth solutions
The power of AI can provide ways to improve the functions and capabilities
of mHealth devices. For example, the capabilities of AI techniques can assist
in building smart solutions that can make mobile devices aware of the needs
pg. 135
and preferences of users by leveraging real-time situational data, such as
location and mood of the users, that is collected via sensors and other data
inputs (e.g., survey apps) (Luxton, June, Sano, & Bickmore, 2016). This similar
method can be adopted in mHealth to help address the accuracy of data,
particularly in solutions where evaluation of the data is limited or where
automation takes precedence in providing feedback to users (Automated Apps,
Text-messaging feedback etc).
AI is a growing field and its use is expanding, particularly where sensor-based
solutions existed. In a study Luxton et al. (2016), Ambient Intelligence (Aml) is
a term that refers to smart technologies that are embedded within a space, such
as in a home, hospital room, or a public area that interact with persons within
the environment. Aml gathers information from sensors and other integration
devices that provide contextual data to adapt environments to users’ needs in
an interactive manner Luxton et al. (2016). The characteristics of Aml are
Luxton et al. (2016):
• Context Aware: It exploits the contextual and situational information.
• Personalized: It is personalized and tailored to the needs of each individual.
• Anticipatory: It can anticipate the needs of an individual without the
conscious mediation of the individual.
• Adaptive: It adapts to the changing needs of individuals.
• Ubiquity: It is embedded and is integrated into our everyday environments.
• Transparency: It recedes into the background of our daily life in an
unobtrusive way.
This technique can be applied to different mHealth devices where algorithms
can be adapted to the specific solutions and check for data quality as it gets
collected, analyzed and disseminated. This provides both accurate data and
help improve the integrity of the data. With the technologies and advances in
Artificial intelligence, earning the trust of the users to use the IT systems is of a
concern and must be taken into consideration as it can deter patients from
pg. 136
adopting technologies. With the promising results of Machine Learning and
Artificial intelligence, the privacy of people must be respected, and the data
must be secure as health data is sensitive data.
5.6 Privacy of users and data security
With the introduction of ML and AI in the healthcare domain, user privacy needs
to be respected and looked after while maintaining the security of their records.
In a world where data is in abundance, it creates an opportunity for algorithms
to be used in detecting a various number of activities for whatever the solution
the data is required for. This level of intelligence can penetrate users’ privacy
and breach their privacy as it generates information that can reveal information
about the users or patients.
The privacy of the users must be respected, and their health data secured to
earn their trust. With data proving to containing more information about the
patients and the diseases, it can penetrate users’ privacy if ML algorithms are
used for the wrong reasons. The intelligence of the machine can prove critical
in overfitting data that can penetrate the users’ privacy as their behaviour is
closely modelled. This can break the trust of users and push them to behave
differently. Privacy can push people to behave differently to what is expected of
them. An example is where between 15% to 17% of US adults have changed
their behaviour in order to protect their privacy by eliminating information from
being disclosed to their medical professionals (Malin, Emam, & O'Keefe, 2013).
Such change in behaviour range from paying out of pocket to avoid disclosure
to not seeking treatment and asking the healthcare professionals to record less
serious health problems (Malin et al., 2013). This can lead to inaccurate,
mismatch in data that impacts the integrity of the information produced because
of this. Evidence suggests that privacy is of concern to people using information
systems (IS) for medical reasons, particularly the use of websites as it impacted
their behaviour (Sampat & Prabhakar, 2017). Those with higher privacy
concerns view IS systems to be more risky and develop concerns about them
(Sampat & Prabhakar, 2017). Another case of protecting users’ privacy is with
the implementation of EHR systems, where the content of the structured and
free-text data means that privacy and security are essential for managing the
pg. 137
systems (Huang, Sharaf & Huang 2013). A modern example of this challenge
is in the implementation of Smart health (s-health), where context-aware
augmentation of mobile health in smart cities provides an opportunity for
accurate and efficient prevention of various diseases and accidents (Zhang,
Zheng, & Deng, 2018). The main concern for users of s-heath is who has
access to their health data, as the expectation of users of the systems is such
that only professional healthcare givers should have access to them, which
presents a limitation in the way access control is granted to people accessing
the systems that can potentially harm the data security (Zhang et al., 2018).
Chapter Summary:
The research was addressed using the J48 Algorithm in Weka Tool. However,
with the use of such technique, data accuracy must be defined and given a
context in order to understand the causes of inaccurate data. The human
behaviour is another factor that must be taken into consideration when
modelling an algorithm to achieve a certain level of data accuracy and to
accurately predict when or not data is accurate. This gives metadata another
important role in helping researchers and people understand what the data
means and how it affects the modelling. With more metadata, the data can be
become more contextualized thus providing a deeper understanding of the
phenomenon under study, which in this instance, would be how the patient
managed their chronic disease (diabetes). However, with data becoming more
contextualized and machine learning performance of predicting accurately, the
privacy of the users and the security of the data become critical to ensure that
users’ privacy is not exploited, and their data misused.
Chapter 6 : Conclusion This chapter presents an overview of the study and provides a review of how
the research question was addressed. mHealth presents an opportunity in
expanding health services and delivering more care through both self-
management and other technologies. This makes data a fundamental part of
mHealth as data is produced by users of the systems (patients, health
pg. 138
professionals, and analytical data) and it is then transformed into information
that can lead to decision making. The delivery of accurate mHealth solutions
can be achieved through the implementation of Machine Learning algorithms
that can be trained to detect inaccurate data.
The research question was addressed using a qualitative approach where
patients’ data were examined in depth and themes were generated that help
understanding how data inaccuracy occurs. The findings revealed that a
classification algorithm (J48) can be implemented in mHealth solutions to
address the issue of data inaccuracy and information integrity.
The following is a summary of the how the research question was addressed,
the contribution to both theory and practice, the limitations presented in the
study and the future research directions.
6.1 Answering the research question
The purpose of this study was to find ways (the ‘How’ part of the research
question) through which Machine learning can detect inaccurate data to prevent
erroneous data from being used in decision making, deliver information with
Integrity and preventing medical errors. The research question posed in the
study was ‘How can Machine Learning be applied in mHealth solutions to
address data accuracy and Information Integrity?’. The research question was
addressed when data was modelled using a classification algorithm that
detected which instances of patients’ data were accurate or inaccurate. This
was accomplished after the manual labelling of the data which resulted in a
model that checked for the quality of data. The Machine Learning model means
that that mHealth can use this model to enhance the element of data quality in
mHealth solutions to further develop the quality of mHealth solutions, remove
the risk of harming patients from false or inaccurate data that could be used in
decision making, and continue to deliver solutions. As the number of chronic
diseases rises and an increasing number of people seeks self-management to
improve their health, this Machine Learning model can help in assisting that the
information generated by mHealth technology is accurate and can be relied
upon for turning the data into information that can be used to help make a great
pg. 139
impact on peoples’ health status and continue to improve the management of
their diseases. In Table 6.1.1., it illustrates how each element of the data quality
was addressed in this study as they played a role in generating the Machine
Learning Model.
Table 6.1.1: Addressing the Element of Data Quality in this study.
Data Quality Element How this element is addressed in this study
Completeness This element is critical in checking the accuracy of the
data. This element was addressed in this study as it
served as the first check for data accuracy by examing if
the datasets contained complete values (data and time
stamps, code, and the reading). This then allowed the data
to be progressed to the next stage of checking the
Currency and Timelinesss of the data.
Currency & Timeliness This element of time was of high importance too. The time
stamps and the timeliness of the data demonstrated the
human behaviour and the patients’ way of managing
diabetes through the day. This was addressed by
checking the timestamps of each reading, then compared
on hourly, daily and weekly basis. This provided an idea
of when data is generated and at what frequency.
Legibility This element did not require much attention as almost all
of the data were readable, except in cases where random
values were present within datasets that could be
translated or used in the analysis of the data.
Consistency This element provided an insight of how and when data
was generated, providing a pattern of consistent and
inconsistent data. The patterns assisted in modelling the
data and the human behaviour (patients taking their
gluocose readings), which were then used to flag accurate
and inaccurate data.
Reliability While some of the data were complete, and legible, yet,
they were not reliable as they did not provide readings that
could be relied upon for decision making. An example of
this included a single reading with no further explanation
pg. 140
of before or after reading decision making (ie: no
continuation of the action taken by patients).
Meaning and
Usefulness
Data that were complete, current, legible, consistent and
can be relied upon, meant that the data provided a
meaning to what the patients were doing and could be
used for further analysis.
Accuracy and Validity After each dataset was examined against all previous
elements, this element provided the final check that
ensured that data were completed, current & timely,
legible, consistent, reliable, meaningful and useful.
Accessbility This element was not applicable in this study as all data
were accessible.
Confidentiality and
Security
This element could not be addressed as the case used in
this study was secondary and did not included a solution
where data can be monitored right from when the data is
generated to when it is delivered to its destination (medical
professional or automated mHealth solution). Hence, this
element was not tested.
6.2 Contribution to Theory and Practice
This study made contributions to both Theory and Practice. The following
highlights the contributions from the study.
6.2.1 Contribution to Theory
The first contribution to theory is within the IoT domain, where the approach
applied in the analysis of the data can be applied to areas of IoT, such as those
demonsteated by Laplante & Laplante (2016). A second area of contribution
within mHealth is Adherence, where Levensky & O’Donohue (2006) stated that,
despite the advancements in the medical field, adherence is still a problem. The
Algorithm used in this study can be applied to a case study of medical
adherence, whereby the algorithm can detect if a patient is adhering to their
medication by analysing if patients’ data is accurate, that is yielding the
expected outcome. This can in turn enhance the patient-clinician interaction by
providing an efficient way of analysing the patients data to deliver effective
solutions (Kelly-Irving et al., 2009). A third area is health promotion, where the
pg. 141
use of Mobile Apps and SMS (lupton 2012) resulted in Big Data, and the
algorithms used in this study can be applied to filter through the data to provide
an overview of the the health promotion program
Another 2 areas where this study contributely largely to, are Big Data, and Data
Quality. Big Data now includes the veracity variable (Sebaa et al., 2018), that
states the reliability and quality of the data. This contribution to the theory of Big
Data, is by demonstrating how veracity can be achieved through the use of
algorithms that can detect the level of data accuracy. This extends to Data
Quality, where errors in data (Olson, 2003) can now be detected through the
use of algorithms, which enhance the quality of data, leading to better outcomes
and results.
6.2.2 Contribution to Practice
In terms of practice, this study can serve as a guide in impelementing data
quality controls to improve areas mentioned by (Olson, 2003) and (Magrabi et
al., 2016). This can in turn improve the quality of care delivered through
mHealth such as those stated by (Mirza, Norris, & Stockdale, 2008). The study
also contributes to the area of Machine Learning and how it can be implemented
in new emerging technology by analysing the human behaviour and
demonstrating the steps in modelling the data. This includes implementing the
algorithms for chronic disease based solutions such as diabetes (Karan et al.,
2012). The contributions are not limited to this study, as the algorithms can be
adopted by researchers and other interested stakeholders to use within their
solutions.
6.3 Limitations
With the findings and the contributions this study made to the fields of theory
and practice, there were limitations present in the study. The study did not have
access to a medical professional who could contribute to establishing an
understanding of how diabetes works and what the reason for the way humans
behaved, or whether the data were accurate (in the context of the diseases).
However, this study did rely on the supplied metadata that explained diabetes
pg. 142
and provided background information about the codes in the datasets that
explained what was happening with the patients. Access to patients was of
limitation as there were no patients involved in this study which meant the study
lacked information that helped understand how or why the data was sent and
what the patients thought of mHealth. Also, the labelling of the data was based
on the definitions listed in the literature review that helped in forming an
understanding of the characters of each element of the data quality.
6.4 Future research directions
Future research direction from this study is to perform similar studies on
different mHealth technologies such as wearable, implantable, text messaging
and other technologies that are considered as mHealth solution. The future
studies should be conducted based on examining the data from the moment
they are generated (ie: typed by the patient) all the way to when it reaches the
medical professional or the intended destination and compare the original data
with the final data that’s used by medical professionals. The studies should also
be extended to the communication component of mHealth solution and
examine the element of timeliness as delays caused by network outages and
drops in packets can also affect the quality of the data. Other studies are also
needed for examining the security aspects of mHealth solutions and how
metadata can affect the privacy of users transmitting data for analysis using the
Cloud technology.
6.5 Lessons Learnt
Accomplishing the original purpose of this study is a major satisfaction.
However, the study brought some lessons for the research including:
• The research methodology needs to be more thorough and should illustrate
the process of the analysis in greater depth.
• As themes emerged during the literature review, the researcher should
attempt at publishing in academic journals to help in highlighting some more
areas that need to be investigated and studied
pg. 143
• Time management is critical to completing the study on time and a more
realistic milestones should be set with expected outcomes pinned against
real deadlines to ensure the study is moving forward.
The above points were crucial to this study and they were revised multiple times
during the research. Upon revision, they helped move the study forward and at
the end accomplish what was set out to be done, by answering the research
question.
Chapter Summary:
The purpose of the study was to accomplish a way in which Machine Learning
can be applied in mHealth solutions that can detect inaccurate data and achieve
information integrity. The research question was addressed using a J48
classification Algorithm. The results from the study contributed to both Theory
and Practice areas of Information Systems. The research however, did have
some limitations which were highlighted, and lessons were learnt during the
study.
References
Ade, B. (2012). Mobile Healthcare: Challenges and Opportunities Mobile Technology Consumption: Opportunities and Challenges (pp. 77-98). Hershey, PA, USA: IGI Global.
Adeyi, O., Smith, O., & Robles, S. (2007). Public Policy & the Challenge of Chronic Noncommunicable Diseases. Herndon, UNITED STATES: World Bank Publications.
Adibi, S. (2013, 3-7 July 2013). Mobile health (mHealth) biomedical imaging paradigm. Paper presented at the 2013 35th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). Osaka, Japan.
Albert, M. V., Kording, K., Herrmann, M., & Jayaraman, A. (2012). Fall Classification by Machine Learning Using Mobile Phones. PLoS ONE, 7(5), e36556. doi:10.1371/journal.pone.0036556
Amazon. (22/05/2018). AWS IoT. Retrieved from https://aws.amazon.com/iot/
Anpeng, H., Chao, C., Kaigui, B., Xiaohui, D., Min, C., Hongqiao, G., . . . Linzhen, X. (2014). WE-CARE: An Intelligent Mobile Telecardiology System to Enable mHealth Applications. Biomedical and Health Informatics, IEEE Journal of, 18(2), 693-702. doi:10.1109/JBHI.2013.2279136
pg. 144
Ardies, C. M. (2014). Diet, Exercise, and Chronic Disease : The Biological Basis of Prevention (pp. 1). Baton Rouge, UNITED STATES: Chapman and Hall/CRC.
Armstrong, B. K., Gillespie, J. A., Leeder, S. R., Rubin, G. L., & Russell, L. M. (2007). Challenges in health and health care for Australia. Medical Journal of Australia, 187(9), 485-489.
Arslan, P. (2016). Mobile Technologies as a Support Tool for Health. In P. Arslan (Ed.), Mobile Technologies as a Health Care Tool (pp. 11-41). Cham: Springer International Publishing.
Aspalter, C., Pribadi, K. T., & Gauld, R. (2017). Health Care Systems in Developing Countries in Asia. London, UNITED KINGDOM: Routledge.
Atanasova, S., Kamin, T., & Petrič, G. (2018). The benefits and challenges of online professional-patient interaction: Comparing views between users and health professional moderators in an online health community. Computers in Human Behavior, 83, 106-118. doi:https://doi.org/10.1016/j.chb.2018.01.031.
Australian Government. (06/04/2018). Information and Services: Health. Retrieved from https://www.australia.gov.au/information-and-services/health.
Australian Institute of Health and Welfare. (20/05/2018). Australia’s Health, 2016. Retrieved from https://www.aihw.gov.au/getmedia/9844cefb-7745-4dd8-9ee2-f4d1c3d6a727/19787-AH16.pdf.aspx?inline=true
Australian Institute of Health and Welfare. (20/05/2018). Cardiovascular disease, diabetes and chronic kidney disease: Australian facts: prevalence and incidence. Retrieved from https://www.aihw.gov.au/reports/heart-stroke-vascular-disease/cardiovascular-diabetes-chronic-kidney-prevalence/contents/summary
Australian Institute of Health and Welnness. (06/05/2018). Risk factors contributing to chronic diseases. https://www.aihw.gov.au/getmedia/74121d1b-69ca-4a34-a08a-51b741ea26b2/12724.pdf.aspx?inline=true
Baek, H., Suh, J.-W., Kang, S.-H., Kang, S., Lim, H. T., Hwang, H., & Yoo, S. (2018). Enhancing User Experience Through User Study: Design of an mHealth Tool for Self-Management and Care Engagement of Cardiovascular Disease Patients. JMIR Cardio, 2(1), e3. doi:10.2196/cardio.9000 Baker, S. B., Xiang, W., & Atkinson, I. (2017). Internet of Things for Smart Healthcare: Technologies, Challenges, and Opportunities. IEEE Access, 5, 26521-26544. doi:10.1109/ACCESS.2017.2775180.
Baldwin, S. (2011). HEALTHCARE SYSTEMS AROUND THE WORLD. Global Health (1937-514X), 4(1), 1-13.
Baloch, Z. z. b. f. m. e. p., Shaikh, F. K., & Unar, M. A. (2018). A context-aware data fusion approach for health-IoT. International Journal of Information Technology, 10(3), 241-245. doi:10.1007/s41870-018-0116-1.
Bardosh, K. L., Murray, M., Khaemba, A. M., Smillie, K., & Lester, R. (2017). Operationalizing mHealth to improve patient care: a qualitative implementation science evaluation of the WelTel texting intervention in Canada and Kenya. Globalization & Health, 13, 1-15. doi:10.1186/s12992-017-0311-z.
pg. 145
Barjis, J., Kolfschoten, G., & Maritz, J. (2013). A sustainable and affordable support system for rural healthcare delivery. Decision Support Systems, 56, 223-233. doi:https://doi.org/10.1016/j.dss.2013.06.005.
Batini, C. a., & Scannapieco, M. (2016). Data and information quality : dimensions, principles and techniques: Switzerland : Springer, [2016].
Bell, J. (2014). Machine learning: hands-on for developers and technical professionals.
Bengtsson, T. (2010). Introduction. In T. Bengtsson (Ed.), Population Ageing - A Threat to the Welfare State? The Case of Sweden (pp. 1-6). Berlin, Heidelberg: Springer Berlin Heidelberg.
Beratarrechea, A., Lee, A. G., Willner, J. M., Jahangir, E., Ciapponi, A., & Rubinstein, A. (2014). The impact of mobile health interventions on chronic disease outcomes in developing countries: a systematic review. Telemedicine and e-Health, 20(1), 75-82. Bert, F., Giacometti, M., Gualano, M. R., & Siliquini, R. (2013). Smartphones and Health Promotion: A Review of the Evidence. Journal of Medical Systems, 38(1), 9995. doi:10.1007/s10916-013-9995-7
Bigi, S. (2016). Communicating (with) Care : A Linguistic Approach to the Study of Doctor-Patient Interactions. Amsterdam, NETHERLANDS: IOS Press.
Bitsaki, M., Koutras, C., Koutras, G., Leymann, F., Steimle, F., Wagner, S., & Wieland, M. (2017). ChronicOnline: Implementing a mHealth solution for monitoring and early alerting in chronic obstructive pulmonary disease. Health Informatics Journal, 23(3), 197-207. doi:10.1177/1460458216641480 Boccolini, P. d. M. M., Duarte, C. M. R., Marcelino, M. A., & Boccolini, C. S. (2017). Desigualdades sociais nas limitações causadas por doenças crônicas e deficiências no Brasil: Pesquisa Nacional de Saúde 2013. Ciência & Saúde Coletiva, 22, 3537-3546.
Boslaugh, S., & Boslaugh, S. (2007). An Introduction to Secondary Data Analysis Secondary Data Sources for Public Health: Cambridge University Press.
Boulos, M. N. K., Brewer, A. C., Karimkhani, C., Buller, D. B., & Dellavalle, R. P. (2014). Mobile medical and health apps: state of the art, concerns, regulatory control and certification. Online Journal of Public Health Informatics, 5(3), 229-229. doi:10.5210/ojphi.v5i3.4814 Boulos, M. N. K., Wheeler, S., Tavares, C., & Jones, R. (2011). How smartphones are changing the face of mobile and participatory healthcare: an overview, with example from eCAALYX. BioMedical Engineering OnLine, 10(1), 24-37. doi:10.1186/1475-925X-10-24
Bovell-Benjamin, A. (2016). Chronic Diseases : The Escalating Dilemma in Developing Countries. New York: Nova Science Publishers, Inc.
Bowers, D., House, A., Owens, D., Bowers, D., & Owens, D. (2011). Getting Started in Health Research. Hoboken Hoboken, NJ: Wiley.
Bowling, A. (2009). Research Methods in Health: Investigating Health and Health Services (3rd ed ed.). Maidenhead: Maidenhead : McGraw-Hill Education.
pg. 146
Bowman, S. (2013). Impact of Electronic Health Record Systems on Information Integrity: Quality and Safety Implications. Perspectives in Health Information Management, 1-19 19p.
Brunner-La Rocca, H.-P., Fleischhacker, L., Golubnitschaja, O., Heemskerk, F., Helms, T., Hoedemakers, T., . . . Ramaekers, J. (2016). Challenges in personalised management of chronic diseases—heart failure as prominent example to advance the care process. EPMA Journal, 7(1), 2.
Cafazzo, J. A., Casselman, M., Hamming, N., Katzman, D. K., & Palmert, M. R. (2012). Design of an mHealth app for the self-management of adolescent type 1 diabetes: a pilot study. Journal of Medical Internet Research, 14(3).
Carlisle, K., Fleming, R., & Berrigan, A. (2016). Commissioning for healthcare: a case study of the general practitioners After Hours Program. Australian Journal of Primary Health, 22(1), 22-25. doi:/10.1071/PY15036.
Caronna, C. A. (2010). 4. Why Use Qualitative Methods to Study Health Care Organizations? Insights from Multi-Level Case Studies. The SAGE Handbook Qualitative Methods Health Research. SAGE Publications Ltd. London: SAGE Publications Ltd.
Catarinucci, L., De Donno, D., Mainetti, L., Palano, L., Patrono, L., Stefanizzi, M. L., & Tarricone, L. (2015). An IoT-Aware Architecture for Smart Healthcare Systems. IEEE Internet of Things Journal, 2(6), 515-526. doi:10.1109/JIOT.2015.2417684.
Centers for Disease Control and Prevention. National diabetes statistics report. 2014. http://www.cdc.gov/diabetes/pubs/statsreport14/national-diabetes-report-web.pdf. Accessed 1 Jan 2016.
Chandhanayingyong, C., Tangtrakulwanich, B., & Kiriratnikom, T. (2007). Teleconsultation for emergency orthopaedic patients using the multimedia messaging service via mobile phones. Journal of Telemedicine and Telecare, 13(4), 193-196.
Charness, N., Demiris, G., & Krupinski, E. (2011). Designing Telehealth for an Aging Population : A Human Factors Perspective. Baton Rouge, UNITED STATES: Chapman and Hall/CRC.
Chiarini, G., Ray, P., Akter, S., Masella, C., & Ganz, A. (2013). mHealth Technologies for Chronic Diseases and Elders: A Systematic Review. IEEE Journal on Selected Areas in Communications, 31(9), 6-18. doi:10.1109/JSAC.2013.SUP.0513001 Chib, A., Car, J., & Van Velthoven, M. H. (2015). MHealth adoption in low-resource environments: A review of the use of mobile healthcare in developing countries. Journal of Health Communication, 20(1), 4-34. doi:10.1080/10810730.2013.864735.
Chow, C. K., Ariyarathna, N., Islam, S. M. S., Thiagalingam, A., & Redfern, J. (2016). mHealth in Cardiovascular Health Care. Heart, Lung and Circulation, 25(8), 802-807. doi:https://doi.org/10.1016/j.hlc.2016.04.009.
Coady, D., Clements, B. J., Gupta, S., & International Monetary, F. (2012). Economics of Public Health Care Reform in Advanced and Emerging Economies (pp. 3) Washington, DC, UNITED STATES: International Monetary Fund.
pg. 147
Cohen, M. Z., Steeves, R. H., & Kahn, D. L. (2000). Hermeneutic Phenomenological Research : A Practical Guide for Nurse Researchers. Thousand Oaks, Calif: SAGE Publications, Inc.
Committee on Oral Health Access to, S., Board on Health Care, S., Board on Children, Y., Families, Institute of, M., Division of, B., . . . National Research, C. (2011). Improving Access to Oral Health Care for Vulnerable and Underserved Populations. Washington, UNITED STATES: National Academies Press.
Corbin, J. M., & Strauss, A. L. (2008). Basics of qualitative research : techniques and procedures for developing grounded theory: Los Angeles, Calif. ; London : SAGE, 2008.3e [ed.] / Juliet Corbin, Anselm Strauss.
Creswell, J. W. (2009). Research design : qualitative, quantitative, and mixed methods approaches: Thousand Oaks, Calif. : Sage Publications, c2009.3rd ed.
Creswell, J. W. a. (2014). Research design : qualitative, quantitative, and mixed methods approaches: Thousand Oaks, California SAGE Publications, Inc., 2014. Fourth edition, international student edition.
Cucoranu, I. C., Parwani, A. V., West, A. J., Romero-Lauro, G., Nauman, K., Carter, A. B., . . . Pantanowitz, L. (2013). Privacy and security of patient data in the pathology laboratory. J Pathol Inform, 4(1), 23-39. doi:10.4103/2153-3539.108542
Cunningham, P. (2012). It's most important role: ensuring information integrity, 20.
Curioso, W. H., & Mechael, P. N. (2010). Enhancing 'M-health' with south-to-south collaborations. Health affairs (Project Hope), 29(2), 264. doi:10.1377/hlthaff.2009.1057.
Dambha-Miller, H., Silarova, B., Irving, G., Kinmonth, A. L., & Griffin, S. J. (2017). Patients’ views on interactions with practitioners for type 2 diabetes: a longitudinal qualitative study in primary care over 10 years (pp. 1). British Journal of General Practice.
Datskovsky, G., Hedges, R., Empel, S., & Washington, L. (2015). Evaluating the Information Governance Principles for Healthcare: Integrity and Protection. Journal of AHIMA, 86(4), 48-49.
David, W., Sara, U., & Erin, J. (2011). Personal Health Information in the Age of Ubiquitous Health Clinical Technologies: Concepts, Methodologies, Tools and Applications (pp. 50-72). Hershey, PA, USA: IGI Global
De Mauro, A., Greco, M., & Grimaldi, M. (2015). What is Big Data? A Consensual Definition and a Review of Key Research Topics. AIP Conference Proceedings, 1644(1), 97-104. doi:10.1063/1.4907823.
De Vries, A., de Roten, Y., Meystre, C., Passchier, J., Despland, J. N., & Stiefel, F. (2014). Clinician characteristics, communication, and patient outcome in oncology: a systematic review. Psycho‐Oncology, 23(4), 375-381.
Denzin, N. K., & Lincoln, Y. S. (2011). The Sage handbook of qualitative research: Thousand Oaks : Sage, c2011.4th ed.
Djam, X., & Kimbi, Y. (2011). Fuzzy expert system for the management of hypertension. The Pacific Journal of Science and Technology, 12(1), 390-402.
pg. 148
Dobson, R., Carter, K., Cutfield, R., Hulme, A., Hulme, R., McNamara, C., . . . Strydom, J. (2015). Diabetes text-message self-management support program (SMS4BG): a pilot study. JMIR mHealth and uHealth, 3(1).
Dolley, S. (2018). Big Data's Role in Precision Public Health. Frontiers in Public Health.
Donley, A. M. (2012). Research Methods. New York: Infobase Publishing.
Dua, D. and Karra Taniskidou, E. (2017). UCI Machine Learning Repository [http://archive.ics.uci.edu/ml]. Irvine, CA: University of California, School of Information and Computer Science.
Duckett, S., & Willcox, S. (2015). Australian Health Care System. Sydney, AUSTRALIA: Oxford University Press.
Dugas, M., Crowley, K., Gao, G. G., Xu, T., Agarwal, R., Kruglanski, A. W., & Steinle, N. (2018). Individual differences in regulatory mode moderate the effectiveness of a pilot mHealth trial for diabetes management among older veterans. PLoS ONE(3). doi:10.1371/journal.pone.0192807
Dumas, M. B. (2013). Diving Into the Bitstream : Information Technology Meets Society in a Digital World. New York: Routledge.
Dyas, A. R., et al. (2018). "Reducing cost and improving operating room efficiency: examination of surgical instrument processing." Journal of Surgical Research 229: 15-19.
Eisele, T. P., Silumbe, K., Yukich, J., Hamainza, B., Keating, J., Bennett, A., & Miller, J. M. (2013). Measuring Coverage in MNCH: Accuracy of Measuring Diagnosis and Treatment of Childhood Malaria from Household Surveys in Zambia. PLoS Medicine, 10(5), e1001417. doi:10.1371/journal.pmed.1001417.
Emmel, N. (2013). Sampling and choosing cases in qualitative research: A realist approach: (pp. 37). Sage. Estrin, D., & Sim, I. (2010). Health care delivery. Open mHealth architecture: an engine for health care innovation. Science, 330(6005), 759-760. doi:10.1126/science.1196187
Fadlalla, A., & Wickramasinghe, N. (2004). An integrative framework for HIPAA-compliant I*IQ healthcare information systems. Int J Health Care Qual Assur Inc Leadersh Health Serv, 17(2-3), 65-74
Farahani, B., Firouzi, F., Chang, V., Badaroglu, M., Constant, N., & Mankodiya, K. (2018). Towards fog-driven IoT eHealth: Promises and challenges of IoT in medicine and healthcare. Future Generation Computer Systems, 78, 659-676.
Fawcett, J., & Garity, J. (2009). Chapter 8: evaluation of samples (pp. 133-152 120p). Philadelphia, Pennsylvania: F.A. Davis Company/Publishers.
Feinberg, L., Menon, J., Smith, R., Rajeev, J. G., Kumar, R. K., & Banerjee, A. (2017). Potential for mobile health (mHealth) prevention of cardiovascular diseases in Kerala: A population-based survey. Indian Heart Journal, 69(2), 182-199. doi:https://doi.org/10.1016/j.ihj.2016.11.004
Firouzi, F., Rahmani, A. M., Mankodiya, K., Badaroglu, M., Merrett, G. V., Wong, P., & Farahani, B. (2018). Internet-of-Things and big data for smarter healthcare: From
pg. 149
device to architecture, applications and analytics. Future Generation Computer Systems, 78, 583-586. doi:https://doi.org/10.1016/j.future.2017.09.016
Flickinger, T. E., et al. (2016). "Clinician empathy is associated with differences in patient-clinician communication behaviors and higher medication self-efficacy in HIV care." Patient education and counseling 99(2): 220-226.
Flowerday, S., & Solms, R. v. (2010). What constitutes information integrity? South African Journal of Information Management(2).
Fortmann, A. L., Garcia, M. I., Taleb, M., Euyoque, J. A., Clark, T., Skidmore, J., . . . Schultz, J. (2017). Dulce digital: An mHealth SMS based intervention improves glycemic control in hispanics with type 2 diabetes. Diabetes Care, 40(10), 1349-1355. doi:10.2337/dc17-0230.
Foscolou, A., Magriplis, E., Tyrovolas, S., Soulis, G., Bountziouka, V., Mariolis, A., . . . Panagiotakos, D. (2018). Lifestyle determinants of healthy ageing in a Mediterranean population: The multinational MEDIS study. Experimental Gerontology, 110, 35-41. doi:https://doi.org/10.1016/j.exger.2018.05.008.
Fox, S., & Duggan, M. (2012). Mobile health 2012. Washington, DC: Pew Internet & American Life Project.
Gabriel, J. (2017). Understanding the challenges to vascular access in an ageing population. British Journal of Nursing, 26(14), S15-S23.
Gao, X., Chen, Z., Tang, S., Zhang, Y., & Li, J. (2016). Adaptive weighted imbalance learning with application to abnormal activity recognition. Neurocomputing, 173, Part 3, 1927-1935. doi:http://dx.doi.org/10.1016/j.neucom.2015.09.064.
Gaugler, J. E. (2015). Chronic Disease and Aging:Screening and Management Applications. Journal of Applied Gerontology, 34(3), 273-276. doi:10.1177/0733464815573392.
Georgsson, M., & Staggers, N. (2016). An evaluation of patients' experienced usability of a diabetes mHealth system using a multi-method approach. Journal of Biomedical Informatics, 115. doi:10.1016/j.jbi.2015.11.008.
Gideon, L. (2012). Handbook of Survey Methodology for the Social Sciences. New York, NY: Springer New York: New York, NY.
Gillham, B. (2010). Case Study Research Methods. London, GBR: Continuum International Publishing.
Gleason, A. W. (2015). mHealth — Opportunities for Transforming Global Health Care and Barriers to Adoption. Journal of Electronic Resources in Medical Libraries, 12(2), 114-125. doi:10.1080/15424065.2015.1035565.
Glowik, M., & Smyczek, S. (2015). Healthcare : Market Dynamics, Policies and Strategies in Europe. Berlin, GERMANY: De Gruyter.
Goertz, C. M., Salsbury, S. A., Long, C. R., Vining, R. D., Andresen, A. A., Hondras, M. A., . . . Wallace, R. B. (2017). Patient-centered professional practice models for managing low back pain in older adults: a pilot randomized controlled trial. BMC Geriatrics, 17(1), 235. doi:10.1186/s12877-017-0624-z.
pg. 150
Goggin, G. (2008). Mobile Phone Cultures. London, UNITED KINGDOM: Taylor & Francis Group. González-Ferrer, A. a. g. c., Seara, G., Cháfer, J., & Mayol, J. (2018). Generating Big Data Sets from Knowledge-based Decision Support Systems to Pursue Value-based Healthcare. International Journal of Interactive Multimedia & Artificial Intelligence, 4(7), 42-46. doi:10.9781/ijimai.2017.03.006.
Grbich, C. (2013). Qualitative data analysis : an introduction: London ; Thousand Oaks, Calif. : SAGE, 2013.2nd ed.
Gregori, D., & Berchialla, P. (2012). Quality of Electronic Medical Records Statistical Methods in Healthcare (pp. 456-480): John Wiley & Sons, Ltd.
Griffin, P. M., Black Nembhard, H., Deflitch, C., Bastian, N. D., Kang, H., & Munoz, D. A. (2016). Healthcare Systems Engineering. Hoboken, UNITED STATES: John Wiley & Sons, Incorporated.
Griffiths, F. (2009). 7 Collecting and Exploring New Data Using Qualitative Methods. Research Methods for Health Care Practice. SAGE Publications, Ltd. London, England: SAGE Publications, Ltd.
Grossmann, C., Goolsby, W. A., Olsen, L., Goolsby, W. A., Olsen, L., & McGinnis, J. M. (2011). Engineering a Learning Healthcare System : A Look at the Future: Workshop Summary. Washington, UNITED STATES: National Academies Press.
Gupta, A., & Sharda, R. (2013). Improving the science of healthcare delivery and informatics using modeling approaches. Decision Support Systems, 55(2), 423-427. doi:https://doi.org/10.1016/j.dss.2012.10.001.
Haddad, D., Selsky, J., Hoffman, J. E., Kravitz, R. L., Estrin, D. E., & Sim, I. (2012). Making Sense of Mobile Health Data: An Open Architecture to Improve Individual-and Population-Level Health. Journal of Medical Internet Research, 14(4), 10-10.
Halpert, A. (2018). Irritable Bowel Syndrome: Patient-Provider Interaction and Patient Education . Journal of Clinical Medicine, 7(1), 1-10. doi:10.3390/jcm7010003.
Hamel, M. B., Cortez, N. G., Cohen, I. G., & Kesselheim, A. S. (2014). FDA Regulation of Mobile Health Technologies. The New England Journal of Medicine, 371(4), 372-379.
Hamine, S., Gerth-Guyette, E. e. p. o., Faulx, D., Green, B. B., & Ginsburg, A. S. (2015). Impact of mHealth Chronic Disease Management on Treatment Adherence and Patient Outcomes: A Systematic Review. Journal of Medical Internet Research, 17(2), 7-7. doi:10.2196/jmir.3951.
Harris, M. F., & Zwar, N. A. (2007). Care of patients with chronic disease: the challenge for general practice. Medical Journal of Australia, 187(2), 104.
Hayes, B. M., & Aspray, W. (2010). Health Informatics. Cambridge, UNITED STATES: MIT Press.
Healey, B. J., & Evans, T. M. (2015). Introduction to health care services : foundations and challenges: (pp. 299). San Francisco, CA : Jossey-Bass A Wiley Brand, [2015].
healthdirectAustralia. (20/05/2018). Equity access for all Australians. Retrieved from https://about.healthdirect.gov.au/equity-of-access-for-all-australians
pg. 151
Health informatics : improving patient care. (2012). Swindon, U.K. : British Informatics Society Ltd., 2012.
Henriquez-Camacho, C., Losa, J., Miranda, J. J., & Cheyne, N. E. (2014). Addressing healthy aging populations in developing countries: unlocking the opportunity of eHealth and mHealth. Emerging Themes in Epidemiology, 11(1), 14-29. doi:10.1186/s12982-014-0021-4
Heszen-Klemens, I., & Lapińska, E. (1984). Doctor-patient interaction, patients' health behavior and effects of treatment. Social Science & Medicine, 19(1), 9-18. doi:https://doi.org/10.1016/0277-9536(84)90132-1.
Hinkson, I. V., Davidsen, T. M., Klemm, J. D., Chandramouliswaran, I., Kerlavage, A. R., & Kibbe, W. A. (2017). A Comprehensive Infrastructure for Big Data in Cancer Research: Accelerating Cancer Research and Precision Medicine. Frontiers in Cell and Developmental Biology, 5(83). doi:10.3389/fcell.2017.00083.
Hoffer, E. R. a. (2016). Designing interconnected (IoT) environments: Sebastopol, CA: O'Reilly, [2016].
Hosseini, H. (2015). Aging and the Rising Costs of Healthcare in the United States: Can There be a Solution? Ageing International, 40(3), 229-247.
Howdon, D., & Rice, N. (2018). Health care expenditures, age, proximity to death and morbidity: Implications for an ageing population. Journal of Health Economics, 57, 60-74. doi:https://doi.org/10.1016/j.jhealeco.2017.11.001.
Huang, T., Lan, L., Fang, X., An, P., Min, J., & Wang, F. (2015). Promises and Challenges of Big Data Computing in Health Sciences. Big Data Research, 2(1), 2-11. doi:https://doi.org/10.1016/j.bdr.2015.02.002.
Hussain, F. (2017). Internet of Everything. In Internet of Things: Building Blocks and Business Models (pp. 1-11). Cham: Springer International Publishing.
IBM. (22/05/2018). Watson Internet of Things. Retrieved from https://www.ibm.com/internet-of-things
International Business Machines. (05/05/2018). Watson. Retrieved from https://www.ibm.com/watson/health/value-based-care/enable-effective-care/
International Telecommunication Union. (04/04/2018). Key ICT Indicators for developed and developing countries and the world (totals and penetration rates (2015).Retrieved from http://www.itu.int/en/ITUD/Statistics/Documents/statistics/2015/ITU_Key_2005-2015_ICT_data.xls
Islam, S. M. R., Kwak, D., Kabir, M. H., Hossain, M., & Kwak, K. S. (2015). The Internet of Things for Health Care: A Comprehensive Survey. IEEE Access, 3, 678-708. doi:10.1109/ACCESS.2015.2437951
Istepanian, R. S. H., Hu, S., Philip, N. Y., & Sungoor, A. (2011, Aug. 30 2011-Sept. 3 2011). The potential of Internet of m-health Things “m-IoT” for non-invasive glucose level sensing. Paper presented at the 2011 Annual International Conference of the IEEE Engineering in Medicine and Biology Society. Boston, MA, USA.
pg. 152
Runciman, B., & Gordon, K. (2014). Big Data: Opportunities and challenges. Swindon, UK: BCS, The Chartered Institute for IT.
Jara, A. j. u. e., Zamora, M. m. u. e., & Skarmeta, A. s. u. e. (2014). Drug identification and interaction checker based on IoT to minimize adverse drug reactions and improve drug compliance. Personal & Ubiquitous Computing, 18(1), 5-17. doi:10.1007/s00779-012-0622-2
Jayasena, R., Cellar, B., Sparks, R., Varnfield, M., Li, J., & Nepal, S. (2016). Monitoring of Chronic Disease in the community: Australian Telehealth Study on Organisational Challenges and Economic Impact. International Journal of Integrated Care (IJIC), 16(6), 1-3. doi:10.5334/ijic.2654
Jebadurai, J., & Dinesh Peter, J. (2018). Super-resolution of retinal images using multi-kernel SVR for IoT healthcare applications. Future Generation Computer Systems, 83, 338-346. doi:https://doi.org/10.1016/j.future.2018.01.058
Jenicek, M. (2010). Medical Error and Harm Understanding, Prevention, and Control. Hoboken: Taylor and Francis.
John Hopkins. (04/04/2018). Study Suggests Medical Errors Now Third Leading Cause of Death in the U.S. Retrieved from https://www.hopkinsmedicine.org/news/media/releases/study_suggests_medical_errors_now_third_leading_cause_of_death_in_the_us
Jones, L. K., Pulk, R., Gionfriddo, M. R., Evans, M. A., & Parry, D. (2018). Utilizing big data to provide better health at lower cost. American Journal of Health-System Pharmacy, 75(7), 427-435.
Jusoh, S. s. j. n. e. s. (2017). A Survey on Trend, Opportunities and Challenges of mHealth Apps. International Journal of Interactive Mobile Technologies, 11(6), 73-85. doi:10.3991/ijim.v11i6.7265
Kahn, J. G., Yang, J. S., & Kahn, J. S. (2010). ‘Mobile’health needs and opportunities in developing countries. Health Affairs, 29(2), 252-258.
Kalra, J. (2011). Medical Errors and Patient Safety : Strategies to Reduce and Disclose Medical Errors and Improve Patient Safety. Berlin: De Gruyter.
Kamath, D. U., & Choppella, K. (2017). Mastering Java Machine Learning. Birmingham, UNKNOWN: Packt Publishing.
Karan, O., Bayraktar, C., Gümüşkaya, H., & Karlık, B. (2012). Diagnosing diabetes using neural networks on small mobile devices. Expert Systems with Applications, 39(1), 54-60. doi:http://dx.doi.org/10.1016/j.eswa.2011.06.046.
Kelly-Irving, M., Rolland, C., Afrite, A., Cases, C., Dourgnon, P., Lombrail, P., . . . Lang, T. (2009). Patient-physician interaction in general practice and health inequalities in a multidisciplinary study: design, methods and feasibility in the French INTERMEDE study (pp. 1). BMC Health Services Research, 9(1), 66. doi:10.1186/1472-6963-9-66.
Kelsey, R. (2016). Patient Safety : Investigating and Reporting Serious Clinical Incidents. Boca Raton, UNITED KINGDOM: CRC Press.
Kendig, H., McDonald, P., & Piggott, J. (2016). Introduction.
pg. 153
A multidisciplinary approach to ageing. In Population Ageing and Australia's Future (pp. 3-18): ANU Press.
Kennedy, B. K., Berger, S. L., Brunet, A., Campisi, J., Cuervo, A. M., Epel, E. S., . . . Sierra, F. (2014). Geroscience: Linking Aging to Chronic Disease. Cell, 159(4), 709-713. doi:https://doi.org/10.1016/j.cell.2014.10.039.
Kim, S., & Kim, S. (2018). User preference for an IoT healthcare application for lifestyle disease management. Telecommunications Policy, 42(4), 304-314. doi:https://doi.org/10.1016/j.telpol.2017.03.006.
Kim, Y. S. (2013). Biblical Interpretation : Theory, Process, and Criteria. Eugene, Or: Pickwick Publications.
Kitsiou, S., Paré, G., Jaana, M., & Gerber, B. (2017). Effectiveness of mHealth interventions for patients with diabetes: an overview of systematic reviews. PLoS ONE, 12(3), e0173160.
Klein, L. A., & Neumann, E. L. (2008). Integrated Health Care Delivery. New York, UNITED STATES: Nova Science Publishers, Incorporated.
Klemettinen, M. (2007). Introduction Enabling Technologies for Mobile Services (pp. 1-5): John Wiley & Sons, Ltd. Klonoff, D. C. (2013). The current status of mHealth for diabetes: will it be the next big thing? J Diabetes Sci Technol, 7(3), 749-758.
Knickman, J., Kovner, A. R., & Jonas, S. (2015). Jonas and Kovner's Health Care Delivery in the United States, 11th Edition (Vol. 11th edition). New York, NY: Springer Publishing Company.
Kohn, L. T., Corrigan, J., & Donaldson, M. S. (2000). To Err Is Human : Building a Safer Health System. Washington, D.C.: National Academies Press.
Komp, K., & Johansson, S. (2015). Introduction. In Population ageing from a lifecourse perspective (1 ed., pp. 1-10): Policy Press at the University of Bristol.
Konschak, C., & Jarrell, L. (2010). Consumer-Centric Healthcare : Opportunities and Challenges for Providers. Chicago, IL, USA: Health Administration Press.
Konschak, C., Nayyar, G., Morris, W. H., & Levin, D. (2013). MHealth : Global Opportunities and Challenges. Virginia Beach, VA: Convurgent Publishing.
Koumpouros, Y. (2014). Big data in healthcare. Healthcare Administration: Concepts, Methodologies, Tools, and Applications: Concepts, Methodologies, Tools, and Applications, 23.
Kreps, G. L. (2017). The relevance of health literacy to mHealth. Information Services and Use, 37(2), 123-130. doi:10.3233/ISU-170828
Kumar, S., Nilsen, W. J., Abernethy, A., Atienza, A., Patrick, K., Pavel, M., . . . Swendeman, D. (2013). Mobile Health Technology Evaluation: The mHealth Evidence Workshop. American Journal of Preventive Medicine, 45(2), 228-236. doi:http://dx.doi.org/10.1016/j.amepre.2013.03.017.
Lambin, P., Roelofs, E., Reymen, B., Velazquez, E. R., Buijsen, J., Zegers, C. M., . . . Dekker, A. (2013). 'Rapid Learning health care in oncology' - an approach towards
pg. 154
decision support systems enabling customised radiotherapy'. Radiother Oncol, 109(1), 159-164. doi:10.1016/j.radonc.2013.07.007.
Lantz, B. (2013). Machine learning with R. Packt Publishing Ltd.
Laplante, P. A., & Laplante, N. (2016). The Internet of Things in Healthcare: Potential Applications and Challenges. IT Professional, 18(3), 2-4. doi:10.1109/MITP.2016.42.
Le Bodic, G. (2005). Basic Concepts Mobile Messaging Technologies and Services (pp. 1-27): John Wiley & Sons, Ltd.
LeRouge, C., & Wickramasinghe, N. (2013). A review of user-centered design for diabetes-related consumer health informatics technologies. J Diabetes Sci Technol, 7(4), 1039-1056.
Leslie, R. D. G., Lansang, C., & Coppack, S. (2012). Diabetes : Clinican's Desk Reference. London, UNKNOWN: Manson Publishing Ltd.
Lev-Ram, M. (2012). THE SUPERCOMPUTER WILL SEE YOU NOW. Fortune, 166(9), 50.
Levensky, E. R., & O'Donohue, W. T. (2006). Promoting Treatment Adherence : A Practical Handbook for Health Care Providers (pp. 3). Thousand Oaks, Calif: SAGE Publications, Inc.
Lin, J. Y. (2013). Mobile Health Tracking of Sleep Bruxism for Clinical, Research, and Personal Reflection.
Lin, R., Ye, Z., Wang, H., & Wu, B. (2018). Chronic Diseases and Health Monitoring Big Data: A Survey. IEEE Reviews in Biomedical Engineering, 1-1. doi:10.1109/RBME.2018.2829704.
Linda, L. K. (2012). Information Integrity: A High Risk, High Cost Vulnerability Proper information goverance includes paying attention to some key building blocks.(GOVERNANCE). Health Data Management, 20(4), 44.
Lupton, D. (2012). M-health and health promotion: The digital cyborg and surveillance society. Social Theory & Health, 10(3), 229-244. doi:10.1057/sth.2012.6
Luxton, D. D., June, J. D., Sano, A., & Bickmore, T. (2016). Chapter 6 - Intelligent Mobile, Wearable, and Ambient Technologies for Behavioral Health Care Artificial Intelligence in Behavioral and Mental Health Care (pp. 137-162). San Diego: Academic Press.
Magrabi, F., Ong, M.-s., & Coiera, E. (2016). An Overview of HIT-Related Errors. In A. Agrawal (Ed.), Safety of Health IT: Clinical Case Studies (pp. 11-23). Cham: Springer International Publishing.
Mahmood, N., Burney, A., Abbas, Z., & Rizwan, K. (2012). Data and Knowledge Management in Designing Healthcare Information Systems. growth, 9(10), 11.
Malin, B. A., Emam, K. E., & O'Keefe, C. M. (2013). Biomedical data privacy: problems, perspectives, and recent advances (Vol. 20, pp. 2).
Malvey, D., & Slovensky, D. J. (2014). From Telemedicine to Telehealth to eHealth: Where Does mHealth Fit? In D. Malvey & D. J. Slovensky (Eds.), mHealth: Transforming Healthcare (pp. 19-43). Boston, MA: Springer US.
pg. 155
Marconi, K., & Lehmann, H. (2014). Big Data and Health Analytics. Philadelphia, PA, UNITED STATES: Auerbach Publications.
Marr, B. (2015). Big Data : Using SMART Big Data, Analytics and Metrics to Make Better Decisions and Improve Performance. New York, UNITED KINGDOM: John Wiley & Sons, Incorporated.
Martha, E. P., & Bart, P. (2006). Computer science tools and techniques Pervasive Computing in Healthcare (pp. 21-45): CRC Press.
McGraw, D. (2012). Building public trust in uses of Health Insurance Portability and Accountability Act de-identified data. Journal of the American Medical Informatics Association : JAMIA, 20(1), 29. doi:10.1136/amiajnl-2012-000936.
Medhanyie, A. A., Spigt, M., Yebyo, H., Little, A., Tadesse, K., Dinant, G.-J., & Blanco, R. (2017). Quality of routine health data collected by health workers using smartphone at primary health care in Ethiopia. International Journal of Medical Informatics, 101, 9-14. doi:10.1016/j.ijmedinf.2017.01.016.
Mena, L. J., Felix, V. G., Ostos, R., Gonzalez, J. A., Cervantes, A., Ochoa, A., . . . Maestre, G. E. (2013). Mobile Personal Health System for Ambulatory Blood Pressure Monitoring. 2013, 13. doi:10.1155/2013/598196
Merriam, S. B. (2014). Qualitative Research : A Guide to Design and Implementation (3). Somerset, US: Jossey-Bass.
Microsoft. (22/05/2018). Internet of Things. Retrieved from https://www.microsoft.com/en-au/internet-of-things/healthcare
Minteer, A. a. (2017). Analytics for the Internet of Things (IoT) : intelligent analytics for your intelligent devices: Birmingham, UK : Packt Publishing, 2017.
Mirza, F., Norris, T., & Stockdale, R. (2008). Mobile technologies and the holistic management of chronic diseases. Health Informatics Journal, 14(4), 309-321. Mohammed, M., Khan, M. B., & Bashier, E. B. M. (2016). Machine Learning : Algorithms and Applications. Milton, UNITED KINGDOM: Chapman and Hall/CRC.
Modzelewski, K. L., Stockman, M. C., & Steenkamp, D. W. (2018). Rethinking the Endpoints of mHealth Intervention Research in Diabetes Care. J Diabetes Sci Technol, 12(2), 389-392. doi:10.1177/1932296817733007.
Monda, J., Keipeer, J., & Were, M. C. (2012). Data Integrity Module for Data Quality Assurance Within an e-Health System in Sub-Saharan Africa. Telemedicine Journal and e-Health, 18(1), 5-10. doi:10.1089/tmj.2010.0224.
Monsen, K., Handler, H., Le, S., & Riemer, J. (2014). Feasibility of using the Omaha System for self-report of symptoms by adults with diabetes. Available in the. Online Journal of Nursing Informatics (OJNI), 18(3)
Mora, H., Gil, D., Azorín, J., Terol, R. M., & Szymanski, J. (2017). An IoT-based computational framework for healthcare monitoring in mobile environments. Sensors (Switzerland), 17(10). doi:10.3390/s17102302.
Moretta Tartaglione, A., Cavacece, Y., Russo, G., & Cassia, F. (2018). The excellence of patient-centered healthcare: Investigating the links between empowerment, co-
pg. 156
creation and satisfaction. TQM Journal, 30(2), 153-167. doi:10.1108/TQM-11-2017-0138.
Morewitz, S. J. (2006). Chronic diseases and health care : new trends in diabetes, arthritis, osteoperosis, fibromyalgia, low back pain, cardiovascular disease, and cancer: New York : Springer, c2006.
Mottl, J. (2014). The imperative of safety in mHealth and why it can't be ignored.
Murthy, R., & Kotz, D. (2014). Assessing blood-pressure measurement in tablet-based mHealth apps. Paper presented at the COMSNETS.
Murdoch, T. B., & Detsky, A. S. (2013). The inevitable application of big data to health care. JAMA, 309(13), 1351-1352. doi:10.1001/jama.2013.393.
Murray, K. A., Belanger, A., Devine, L. T., Lane, A., & Condren, M. E. (2017). Emergency department discharge prescription errors in an academic medical center. Baylor University Medical Center Proceedings, 30(2), 143-146.
Nabi, M., Wahid, A., & Kumar, P. (2017). Performance Analysis of Classification Algorithms in Predicting Diabetes. International Journal of Advanced Research in Computer Science, 8(3), 456-461.
Närman, P., Holm, H., Johnson, P., König, J., Chenine, M., & Ekstedt, M. (2011). Data accuracy assessment using enterprise architecture. Enterprise Information Systems, 5(1), 37-58. doi:10.1080/17517575.2010.507878
Newswire, P. R. (2018). Developed vs developing countries: What's the impact on healthcare coverage? In NY-REPORTLINKER: Y.
Ngwainmbi, E. K. (2014). Healthcare Management Strategy, Communication, and Development Challenges and Solutions in Developing Countries. Lanham: Lexington Books.
Niccoli, T., & Partridge, L. (2012). Ageing as a risk factor for disease. Current Biology, 22(17), R741-R752.
Niilo, S., Ilkka, K., & Elina, M. (2006). Mobile and personal health and wellness management systems. In Pervasive Computing in Healthcare (pp. 105-134): CRC Press.
Nousias, S., Lalos, A. S., Arvanitis, G., Moustakas, K., Tsirelis, T., Kikidis, D., . . . Tzovaras, D. (2018). An mHealth System for Monitoring Medication Adherence in Obstructive Respiratory Diseases Using Content Based Audio Classification. IEEE Access, 6, 11871-11882. doi:10.1109/ACCESS.2018.2809611
Nuovo, J. (2010). Chronic Disease Management. Dordrecht: Springer.
O'Loughlin, M., Mills, J., McDermott, R., & Harriss, L. (2017). Review of patient-reported experience within Patient-Centered Medical Homes: insights for Australian Health Care Homes. Australian Journal of Primary Health, 23(5), 429-439. doi:10.1071/PY17063.
Loukides, M., & Lorica, M. (2016). What Is Artificial Intelligence ?. Retrieved from https://www.oreilly.com/data/free/what-is-artificial-intelligence.csp?intcmp=il-data-free-lp-lgen_20160622_new_site_full_what_is_artificial_intelligence_report_link_top_note
pg. 157
O'Riordan, A., & Elton, J. (2016). Healthcare Disrupted : Next Generation Business Models and Strategies. Hoboken, UNITED STATES: John Wiley & Sons, Incorporated.
O'Shea, C. J., McGavigan, A. D., Clark, R. A., Chew, D. P. B., & Ganesan, A. (2017). Mobile health: an emerging technology with implications for global internal medicine. Internal Medicine Journal(6), 616. doi:10.1111/imj.13440
Olson, J. E. (2003). Data Quality The Accuracy Dimension. Burlington: Elsevier Science.
Oquendo, M. A., Baca-Garcia, E., Artes-Rodriguez, A., Perez-Cruz, F., Galfalvy, H. C., Blasco-Fontecilla, H., . . . Duan, N. (2012). Machine learning and data mining: strategies for hypothesis generation. Molecular Psychiatry, 17, 956+.
Organisation for Economic Co-operation and Development. (07/07/2018). THE ECONOMICS OF PATIENT SAFETY: Strengthening a value-based approach to
reducing patient harm at national level. Retrieved from https://www.oecd.org/els/health-systems/The-economics-of-patient-safety-March-2017.pdf.
Our World In Data (08/05/2018). Risk Factors Picture. Retrieved from https://ourworldindata.org/causes-of-death#comparisons-of-risk-factors-of-death
Patel, M. R., Resnicow, K., Lang, I., Kraus, K., & Heisler, M. (2018). Solutions to Address Diabetes-Related Financial Burden and Cost-Related Nonadherence: Results From a Pilot Study. Health Education & Behavior, 45(1), 101-111. doi:10.1177/1090198117704683
Patil, Y. a. (2017). Azure IoT development cookbook : develop and manage robust IoT solutions: Birmingham, UK : Packt Publishing, 2017.
Patnaik, S., Brunskill, E., & Thies, W. (2009, 17-19 April 2009). Evaluating the accuracy of data collection on mobile phones: A study of forms, SMS, and voice. Paper presented at the Information and Communication Technologies and Development (ICTD), 2009 International Conference on.
Peter, Z., Stefan, F., & Markus, M. (1999). Ageing of population and health care expenditure: a red herring? Health Economics, 8(6), 485-496. doi:doi:10.1002/(SICI)1099-1050(199909)8:6<485::AID-HEC461>3.0.CO;2-4.
Phillips, S. M., Cadmus-Bertram, L., Rosenberg, D., Buman, M. P., & Lynch, B. M. (2018). Wearable Technology and Physical Activity in Chronic Disease: Opportunities and Challenges. American Journal of Preventive Medicine, 54(1), 144-150. doi:10.1016/j.amepre.2017.08.015
Pomerantz, J. a. (2015). Metadata: Cambridge, Massachusetts ; London, England : The MIT Press, [2015].
Product Design & Development. Challenges for doctors using fitness trackers & apps. Retrieved February 24, 2015, from http://www.pddnet.com/news/2015/02/challengesdoctorsusingfitnesstrackersapps
PwC. (23/04/2018). Emerging mHealth: Paths for growth. Retrieved from https://www.pwc.com/gx/en/healthcare/mhealth/assets/pwc-emerging-mhealth-full.pdf
pg. 158
Qi, J., Yang, P., Min, G., Amft, O., Dong, F., & Xu, L. (2017). Advanced internet of things for personalised healthcare systems: A survey. Pervasive and Mobile Computing, 41, 132-149. doi:https://doi.org/10.1016/j.pmcj.2017.06.018.
Radoslav, B., & Pavel, S. (2012). Real-Time Robust Heart Rate Estimation Based on Bayesian Framework and Grid Filters Medical Applications of Intelligent Data Analysis: Research Advancements (pp. 67-90). Hershey, PA, USA: IGI Global.
Reese, R. M., Reese, J. L., Kaluza, B., Kamath, D. U., & Choppella, K. (2017). Machine Learning: End-to-End guide for Java developers. Birmingham, UNITED KINGDOM: Packt Publishing.
Rezaeibagha, F., Win, K. T., & Susilo, W. (2015). A systematic literature review on security and privacy of electronic health record systems: technical perspectives. Health Information Management Journal(3), 23. doi:10.12826/18333575.2015.0001.Rezaeibagha.
Rossman, J. (2016). Amazon Way on IoT: [S.l.] : John Rossman, 2016.
Rowland, D. T. (2012). A Silent Revolution. In Population Aging: The Transformation of Societies (pp. 3-17). Dordrecht: Springer Netherlands.
Sadiq, S. e. (2013). Handbook of Data Quality Research and Practice: Springer Berlin Heidelberg : Imprint: Springer.
SafetyandQuality. (13/05/2018). Patient-clinician communication in hospital. Retrieved from https://www.safetyandquality.gov.au/wp-content/uploads/2016/11/Information-sheet-for-healthcare-providers-Improving-patient-clinician-communication.pdf
Sampat, B., & Prabhakar, B. (2017). Privacy Risks and Security Threats in mHealth Apps. Journal of International Technology & Information Management, 26(4), 126-153.
Sanders, J., Powers, B., Grossmann, C., & Institute of, M. (2013). Digital Data Improvement Priorities for Continuous Learning in Health and Health Care : Workshop Summary. Washington, District of Columbia: National Academies Press.
Sannino, G., De Falco, I., & De Pietro, G. (2014). A General-Purpose mHealth System Relying on Knowledge Acquisition through Artificial Intelligence Ambient Intelligence-Software and Applications (pp. 107-115): Springer.
Saxton, J. (2011). Exercise and Chronic Disease : An Evidence-Based Approach. Florence, UNITED STATES: Taylor and Francis.
Schatz, B. R., & Berlin, R. B. (2011). Healthcare infrastructure : health systems for individuals and populations: London : Springer, c2011.
Schiavo, R. (2013). Health Communication : From Theory to Practice. New York, NY, UNITED STATES: John Wiley & Sons, Incorporated.
Seale, C. (2007). Qualitative research practice. [electronic resource]: London : SAGE, 2007.Concise pbk. ed.
Sebaa, A., Chikh, F., Nouicer, A., & Tari, A. (2018). Medical Big Data Warehouse: Architecture and System Design, a Case Study: Improving Healthcare Resources Distribution. Journal of Medical Systems, 42(4), 59. doi:10.1007/s10916-018-0894-9.
pg. 159
Sebastian-Coleman, L. (2013). Measuring Data Quality for Ongoing Improvement : A Data Quality Assessment Framework. Burlington: Morgan Kaufmann.
Serhani, M. A., Harous, S., Navaz, A. N., Menshawy, M. E., & Benharref, A. (2017). New algorithms for processing time-series big EEG data within mobile health monitoring systems. Computer Methods and Programs in Biomedicine, 149, 79-94. doi:10.1016/j.cmpb.2017.07.007.
Sezgin, E., Özkan-Yildirim, S., & Yildirim, S. (2018). Understanding the perception towards using mHealth applications in practice:Physicians’ perspective. Information Development, 34(2), 182-200. doi:10.1177/0266666916684180.
Shaw, J. (2012). Diabetes. [electronic resource] : the silent pandemic and its impact on Australia: Melbourne : Baker IDI Heart & Diabetes Institute, 2012.
Sidani, S., Reeves, S., Hurlock-Chorostecki, C., van Soeren, M., Fox, M., & Collins, L. (2018). Exploring Differences in Patient-Centered Practices among Healthcare Professionals in Acute Care Settings. Health Communication, 33(6), 716-723. doi:10.1080/10410236.2017.1306476
Silverman, D. (2010). Doing qualitative research : a practical handbook: London : SAGE, 2010. 3rd ed.
Singleton, J. F. (1988). Secondary Data Analysis. Journal of Physical Education, Recreation & Dance, 59(4), 38. Sinnett, W. M. (2008). Excellence in Information Integrity. Florham Park, NJ: Financial Executives Research Foundation, Inc.
Slater, H., Campbell, J. M., Stinson, J. N., Burley, M. M., & Briggs, A. M. (2017). End User and Implementer Experiences of mHealth Technologies for Noncommunicable Chronic Disease Management in Young Adults: Systematic Review. Journal of Medical Internet Research, 19(12).
Smith, J. (2011). Healthcare Management. Maidenhead, UNKNOWN: McGraw-Hill Education.
Steinberg, D. M. a. (2015). The social work student's research handbook: Abingdon, Oxon ; New York, NY : Routledge, 2015.Second edition.
Stey, A. M., Brook, R. H., Needleman, J., Hall, B. L., Zingmond, D. S., Lawson, E. H., & Ko, C. Y. (2015). Hospital Costs by Cost Center of Inpatient Hospitalization for Medicare Patients Undergoing Major Abdominal Surgery. Journal of the American College of Surgeons, 220(2), 207-217.e211. doi:10.1016/j.jamcollsurg.2014.10.021.
Stuckler, D., & Siegel, K. (2011). Sick Societies : Responding to the global challenge of chronic disease (pp. 124). Oxford, UNITED KINGDOM: Oxford University Press, Incorporated.
Sun, J., Guo, Y., Wang, X., & Zeng, Q. (2016). mHealth for aging China: opportunities and challenges. Aging and disease, 7(1), 53.
Supervía, M., & López-Jimenez, F. (2018). mHealth and cardiovascular diseases self-management: There is still a long way ahead of us. European Journal of Preventive Cardiology, 0(0), 2047487318766644. doi:10.1177/2047487318766644
pg. 160
Talburt, J. R., & Zhou, Y. (2015). Chapter 11 - ISO Data Quality Standards for Master Data. In Entity Information Life Cycle for Big Data (pp. 191-205). Boston: Morgan Kaufmann.
Tarassenko, L., & Clifton, D. A. (2011). Semiconductor wireless technology for chronic disease management. Electronics Letters, 47(26), S30-S32. doi:10.1049/el.2011.2679.
Taruna, G. (2014). Mobilising Africa: the growing role of mobile phone technology in healthcare. Africa Health, 36(4), 27-29.
Taylor, G. (2013). Mhealth and video conferencing: better care in rural North Carolina. Behavioral healthcare, 33(5), 47-48.
The Omaha System. (04/04/2018). The Omaha System 2005 Chart. Retrieved from http://cmapspublic3.ihmc.us/rid=1290438215218_1896624281_17913/2010-11-22%20Omaha%20System%20for%20NSFr.cmap
Tollmar, K., Bentley, F., & Viedma, C. (2012). Mobile health mashups: Making sense of multiple streams of wellbeing and contextual data for presentation on a mobile device (pp. 65-72).
Tomar, G. S., Chaudhari, N. S., Bhadoria, R. S., & Deka, G. C. (2016). The Human Element of Big Data : Issues, Analytics, and Performance. Boca Raton, UNITED STATES: CRC Press.
Trifu, M. R. t. y. c., & Ivan, M. L. i. g. c. (2014). Big Data: present and future. Database Systems Journal, 5(1), 32-41.
Tupper, C. D. (2011). 3 - Enterprise-Level Data Architecture Practices Data Architecture (pp. 57-73). Boston: Morgan Kaufmann.
United Nations. (04/04/2018). World Population Ageing. Retrieved from http://www.un.org/en/development/desa/population/publications/pdf/ageing/WPA2015_Report.pdf
Van Den Bos, J., Rustagi, K., Gray, T., Halford, M., Ziemkiewicz, E., & Shreve, J. (2011). The $17.1 billion problem: the annual cost of measurable medical errors. Health Affairs, 30(4), 596-603.
van Loggerenberg, F., Vorovchenko, T., & Amirian, P. (2017). Introduction—Improving Healthcare with Big Data. In P. Amirian, T. Lang, & F. van Loggerenberg (Eds.), Big Data in Healthcare: Extracting Knowledge from Point-of-Care Machines (pp. 1-13). Cham: Springer International Publishing.
van Velthoven, M. H., Car, J., Zhang, Y., & Marušić, A. (2013). mHealth series: New ideas for mHealth data collection implementation in low– and middle–income countries. Journal of Global Health, 3(2), 020101. doi:10.7189/jogh.03.020101.
Varshney, U. (2009). Pervasive Healthcare Computing. Dordrecht: Springer.
Vecellio, E., Maley, M. W., Toouli, G., Georgiou, A., & Westbrook, J. (2015). Data quality associated with handwritten laboratory test requests: classification and frequency of data-entry errors for outpatient serology tests. Health Information Management Journal, 44(3), 7-12. doi:10.12826/18333575.2015.0007.Vecellio.
pg. 161
Verma, P., & Sood, S. K. (2018). Cloud-centric IoT based disease diagnosis healthcare framework. Journal of Parallel and Distributed Computing, 116, 27-38. doi:https://doi.org/10.1016/j.jpdc.2017.11.018.
Verzillo, S., Fiorio, C. V., & Gorli, M. (2018). Evaluating organizational change in health care: The patient-centered hospital model (pp. 3). BMC Health Services Research, 18(1), 3. doi:10.1186/s12913-018-2877-4.
Vincent, C. (2010). Patient Safety (2nd ed. ed.). Hoboken: Wiley.
Wall, D. P., Kosmicki, J., DeLuca, T. F., Harstad, E., & Fusaro, V. A. (2012). Use of machine learning to shorten observation-based screening and diagnosis of autism. Translational Psychiatry, 2(4), e100. doi:10.1038/tp.2012.10.
Wang, R. Y. (1996). Beyond accuracy: What data quality means to data consumers. Journal of Management Information Systems, 12(4), 5-34.
Watts, S. (2016). The Internet of Things (IoT) : applications, technology, and privacy issues: New York : Nova Science Publishers, Inc., [2016].
Wei Li, l. e. k. a. k., Cheolwoo Jung, c. e. k. a. k., & Jongtae Park, j. e. k. a. k. (2018). IoT Healthcare Communication System for IEEE 11073 PHD and IHE PCD-01 Integration Using CoAP. KSII Transactions on Internet & Information Systems, 12(4), 1396-1414. doi:10.3837/tiis.2018.04.001
Weiner, J. P. (2012). Doctor-patient communication in the e-health era. Israel Journal of Health Policy Research, 1, 33-33. doi:10.1186/2045-4015-1-33.
Whettem, E. (2012). Diabetes : Diabetes (pp. 8). Welwyn Garden City, UNITED KINGDOM: Routledge.
White, B. K., Burns, S. K., Giglia, R. C., & Scott, J. A. (2016). Designing evaluation plans for health promotion mHealth interventions: A case study of the Milk Man mobile app. Health Promotion Journal of Australia: Official Journal of Australian Association of Health Promotion Professionals(3), 198. White, T. L., & McBurney, D. (2013). Research methods (9th ed. ed.). Belmont, CA: Wadsworth Cengage Learning.
Wickramasinghe, N., & Gururajan, R. (2016). Innovation practice using pervasive mobile technology solutions to improve population health management: a pilot study of gestational diabetes patient care in Australia: Lippincott Williams & Wilkins.
Wickramasinghe, N., Troshani, I., & Goldberg, S. (2010). An Investigation Into The Use Of Pervasive Wireless Technologies To Support Diabetes Self-Care. In A. Coronato, & G. De Pietro (Eds.), Pervasive and Smart Technologies for Healthcare: Ubiquitous Methodologies and Tools (pp. 114-129). Hershey, PA: IGI Global. doi:10.4018/978-1-61520-765-7.ch006.
Wiens, J., & Shenoy, E. S. (2018). Machine Learning for Healthcare: On the Verge of a Major Shift in Healthcare Epidemiology. Clinical Infectious Diseases, 66(1), 149-153. doi:10.1093/cid/cix731.
Willcox, J. C., Campbell, K. J., McCarthy, E. A., Wilkinson, S. A., Lappas, M., Ball, K., . . . Crawford, D. A. (2015). Testing the feasibility of a mobile technology intervention
pg. 162
promoting healthy gestational weight gain in pregnant women (txt4two) - study protocol for a randomised controlled trial. Trials, 16, 209-209. doi:10.1186/s13063-015-0730-1.
Wildenbos, G. A., Peute, L., & Jaspers, M. (2018). Aging barriers influencing mobile health usability for older adults: A literature based framework (MOLD-US). International Journal of Medical Informatics, 114, 66-75. doi:https://doi.org/10.1016/j.ijmedinf.2018.03.012
Willis, E., Reynolds, L., & Keleher, H. (2016). Understanding the Australian Health Care System: Chatswood, N.S.W. : Elsevier Australia, 2016.
3rd edition.
World Health Organization. (04/04/2018). Ageing and health. Retrieved from http://www.who.int/news-room/fact-sheets/detail/ageing-and-health
World Health Organization. (20/05/2018). Cancer. Retrieved from http://www.who.int/en/news-room/fact-sheets/detail/cancer
World Health Organization. (19/05/2018). Cardiovascular diseases (CVDs). Retrieved from http://www.who.int/en/news-room/fact-sheets/detail/cardiovascular-diseases-(cvds)
World Health Organization. (20/05/2018) CONSTITUTION OF THE WORLD HEALTH
ORGANIZATION. Retrieved from http://apps.who.int/gb/bd/PDF/bd47/EN/constitution-en.pdf?ua=1
World Health Organization. (30/04/2018). Diabetes. Retrieved from http://www.who.int/en/news-room/fact-sheets/detail/diabetes
World Health Organization. (04/04/2018). Global Health and Ageing. Retrieved from http://www.who.int/ageing/publications/global_health.pdf
World Health Organization (30/04/2018). GLOBAL REPORT ON DIABETES. Retrieved from http://apps.who.int/iris/bitstream/handle/10665/204871/9789241565257_eng.pdf;jsessionid=79B8179A9942BADC8E8495F9CE84BBF1?sequence=1
World Health Organization. (04/04/2018). GLOBAL STATUS REPROT ON
NONCOMMUNICABLE DISEASES 2014. Retrieved from http://apps.who.int/iris/bitstream/handle/10665/148114/9789241564854_eng.pdf;jsessionid=BC7210F88F73BC4B06EE87692CA5A3B7?sequence=1
World Health Organization. (05/06/2018). Human rights and health. Retrieved from http://www.who.int/news-room/fact-sheets/detail/human-rights-and-health
World Health Organization. (04/04/2018). Integrated chronic disease prevention and
control. Retrieved from http://www.who.int/chp/about/integrated_cd/en/
World Health Organization. (06/04/2018). Health Systems. Retrieved from http://www.who.int/topics/health_systems/en/
Yepes, M., Maurer, J., Viswanathan, B., Gedeon, J., & Bovet, P. (2016). Potential reach of mHealth versus traditional mass media for prevention of chronic diseases: evidence from a nationally representative survey in a middle-income country in Africa. Journal of Medical Internet Research, 18(5).
pg. 163
Yih, Y. (2014). Handbook of Healthcare Delivery Systems. Baton Rouge, UNITED STATES: CRC Press.
Yin, Y., Zeng, Y., Chen, X., & Fan, Y. (2016). The internet of things in healthcare: An overview. Journal of Industrial Information Integration, 1, 3-13. doi:https://doi.org/10.1016/j.jii.2016.03.004
Yin, R. K. (2014). Case study research : design and methods: Los Angeles : SAGE, 2014.Fifth edition.
Yin, R. K. (2016). Qualitative research from start to finish: New York : Guilford Press, [2016] Second Edition.
Youfa, W., Hong, X., Yaqi, H., Lili, H., Dongsong, Z., Wang, Y., . . . Zhang, D. (2017). A Systematic Review of Application and Effectiveness of mHealth Interventions for Obesity and Diabetes Treatment and Self-Management. Advances in Nutrition, 8(3), 449.
Zacher, H. (2014). Patient safety - a psychological perspective : a psychological perspective. Retrieved from https://ebookcentral.proquest.com
Zhang, Y., Zheng, D., & Deng, R. H. (2018). Security and Privacy in Smart Health: Efficient Policy-Hiding Attribute-Based Access Control. IEEE Internet of Things Journal, 5(3), 2130-2145. doi:10.1109/JIOT.2018.2825289.
Zhu, K. Y., Liu, W. D., & Xiao, Y. (2014). Application of Fuzzy Logic Control for Regulation of Glucose Level of Diabetic Patient. In S. Dua, U. R. Acharya, & P. Dua (Eds.), Machine Learning in Healthcare Informatics (Vol. 56, pp. 47-64): Springer Berlin Heidelberg
Zinoviev, D. (2016). Data Science Essentials in Python: Collect - Organize - Explore - Predict - Value: Pragmatic Bookshelf.
pg. 164 pg. 164
Appendix A To view the results and the algorithms from this study, please navigate to the
folder containing the data by copying the below URL and pasting it into a web
browser to view the folders containing the data.
https://1drv.ms/f/s!AhpMjZqGve6kkFyjtvtezY4W-_eQ
Once the URL is opened, you’ll be presented with 4 numbered folders (See
below Screenshot). Folder 1 contains the original data, Folder 2 contains the
new data format, Folder 3 is the folder containing the Weka .ARFF format which
can be imported into WEKA Tool to run the algorithms against, and Folder 4
cotains the actual results from the Algorithms.
The following list of attachments shows how to load data into Weka and using
the J48 Algorithm to classify the data. Note, Folder 3 which contains both
training and testing datasets in the WEKA ARFF Format. They can be loaded
into WEKA without any editing required.
Step 1: Open WEKA and click on Explorer.
pg. 165
Step 2: Once Explorer is opened, click on ‘Open file’
Step 3: After clicking ‘Open file’, navigate to the folder containing the training
and testing datasets in the WEKA ARFF and select a file.
pg. 166
Step 4: Once a file is selected, browse to the next tab ‘Classify’ and under
‘Classifier’ select ‘trees’ and then ‘J48’. Afte the algorithm is selected, press
‘Start’.
pg. 167
Appendix B The following is a list of publications and conference presentations produced
as part of this study.
1. Sako, Z. Z., Adibi, S., & Wickramasinghe, N. (2018). Application of
Hermeneutics in Understanding New Emergin Technologies in Health Care:
An Example from mHealth Case Study. In N. Wickramasinghe, & J. L.
Schafer (Eds.), Theories to Inform Superior Health Informatics Research
and Practice (pp. 29-35). Cham:Springer Internation Publishling.
2. Sako, Z., Adibi. S., and Wickramasinghe, N. (2017), “Understanding new
emerging technologies through Hermeneutics. An example from mHealth”.
Bled eConference June 18-21 2017.
3. Sako, Z. Z., Karpathiou, V., & Wickramasinghe, N. (2016). Data Accuracy
in mHealth. In N. Wickramasinghe, I. Troshani, & J. Tan (Eds.),
Contemporary Consumer Health Informatics (pp. 379-397). Cham: Springer
International Publishing.
4. Sako, Z., Karpathiou, V., Adibi, S., and Wickramasinghe, N. (2016),
“Understanding new emerging technologies through Hermeneutics. An
example from mHealth”. Bled eConference June 19-22 2016.