Upload
others
View
8
Download
0
Embed Size (px)
Citation preview
Innovar para construir el hospital del futuro
Dr. Lluís Blanch Torra
Senior Critical Care
Director Institut d’Investigació I Innovació Parc Taulí I3PT
WFSICCM Council Member
Madrid - May 29, 2019
L.Blanch is inventor of one Corporació Sanitaria Parc
Taulí owned US patent: “Method and system for
managed related patient parameters provided by a
monitoring device,” US Patent No. 12/538,940.
L.Blanch is founder of BetterCare S.L. which is a
research and development company, start up of
Corporació Sanitària Parc Taulí.
Financial Disclosures
Lluis Blanch MD, PhD
Intensive Care Med 2016 DOI 10.1007/s00134-016-4525-y
Computer Driven Intensive Care
Genetic Analysis
“Triage”
Structural & Physiological
Organ Analysis
Organ Reconstruction
and the brain?
Hospital/
Critical Care
of the
Future
3D-printed
organs
Virtual &
BiosensorsNew
Materials
New Disciplines
L. Blanch
Hospital / Critical Care
of the Future
Heterogeneity & Personalization
Data
Information Technologies & Prediction
Humans
Hospital / Critical Care
of the Future
Heterogeneity & Personalization
Data
Information Technologies & Prediction
Humans
THE PRECISION MEDICINE INITIATIVE
Mission statement:
To enable a new era of medicine through research, technology, and policies that empower
patients, researchers, and providers to work together toward development of individualized
care.
- President Obama, January 30, 2015
https://www.whitehouse.gov/precision-medicine
Syndromic Medicine: Guidelines
Precision
Medicine
Personalized Medicine
Protocols for Specific Subgroups
October & December 2018
January to May 2019
Recognized versus Unrecognized ARDS
JAMA. 2016;315(8):788-800
Difficulties of Developing &
Testing Therapies in ALI & ARDS
• ALI/ARDS definitions have deficiencies in
discrimination
• Multiple etiologies: pulmonary versus
extrapulmonary
• Heterogeneous cohorts with different age &
comorbidities: alcohol abuse, diabetes,…
• Multifaceted pathophysiology of pulmonary
injury
8.309.344 Cases of Invasive MV for Nonsurgical
Indications from 1993 to 2009
Journal of Critical Care 2015
139.500
adults/year
in Spain
Mortality ≈ 30%
Hospital / Critical Care
of the Future
Heterogeneity & Personalization
Data
Information Technologies & Prediction
Humans
Sources of Big Data“of course in the ICU”
Iwashyna TJ & Liu V
AnnalsATS 2014;11:1130-5
Iwashyna TJ & Liu V
AnnalsATS 2014;11:1130-5
Continuous
Respiratory Data
Waveforms
Sources of Big Data“of course in the ICU”
Turner MC et al AJRCCM Articles in Press. Published on 17-December-2015
Ware LB et al AJRCCM Articles in Press. Published on 17-December-2015
Long-Term Ozone Exposure & Respiratory Diseases
Critical Care (2017) 21:257
Managing the Scale of Clinical and Biologic Data
Single cell
sequencing
Sampling
multiple organs
Sequencing in
millions of patients
Multiple points per
patient per day
Trillion Data Points
Sep 2018
Supervised learning algorithms to uncover the relationship
between patient clinical features (such as laboratory tests and vital
signs) and mortality in order to predict the outcome in future cases.
Crit Care Med 2017
Area under the receiver operating characteristic
(AUROC) curves for predicting sepsis 4 hr in advance.
Extraction and Machine Learning:
A total of 65 features from the EMR and bedside
monitoring data
Machine learning–based tools embedded
within EMR improve early detection and
prompt treatment in those with early
sepsis and can predict septic shock
Sep 2018
Unsupervised learning algorithms used to uncover naturally
occurring groupings or clusters of patients based on their clinical
characteristics, without targeting a specific outcome
Patient Heterogeneity
Differences in categorical variables
based on phenotype assignment
Differences in the standardized values of each variable by
phenotype on the y-axis, with the individual continuous
variables along the x-axis
ALVEOLI
ARMA
* *
**
JAMA May 19, 2019
Derivation cohort: 20.189 patients with sepsis
Validation cohort: 43.086 patients with sepsis
IL-6
Inflammatory Cytokines Across
Phenotypes
Short-term Mortality by Phenotype
Big Data is Real World Data
Differences with RCTs
Type
Results
Population
Monitoring
Randomization
Cost
Observational
Effectivity
No restrictions
Not Necessary
No
Cheap
Hospital / Critical Care
of the Future
Heterogeneity & Personalization
Data
Information Technologies & Prediction
Humans
Information Technologies & Devices
Information overload
ICU ITs
Lack of Information !
The Connectivity Envelope of the Future ICU
Halpern NA. Chest 2014;145:903
No patient data continuity
Different manufacturers
Different communication protocols
Lack of interoperability
Specific hospital department solutions
ICU & Information: Today
Medical Devices
Capture, integration and
processing system
20
15
Real time webview
HIS
Asynchronies
database
ICU clinical
practice
IEE
DC
Data analytic tools
+offline review
Patient bedside
ICU clinical data
HL7
storage
Waveforms
repository
Research
Monitoring in the Era of
Precision Critical Care
Respir Care 2018;63:464-478
International Journal of Medical Informatics 112 (2018) 166–172
Flow of the data. Relationship between the Clinical
Information System and ICU-DaMa.
Minimum Datasets
& Quality Metrics
The Future
Blanch L et al. Intensive Care Med 2015;41:633-641
Asynchrony index (AI) % per hour in 4 patients
Asynchronies are difficult to predict !!
1.010 p. with ARDS
In 67 ICU MV patients we
analyzed 9,251 h of MV
(9,694,573 breaths)
Double cycling
occurred in 0.6%.
Crit Care Med 2018; 46: 1385–92
Effect of Double Cycling on Tidal Volume
67 patients, 9,251 hours, 9,694,573 breaths
↑↑Strain
↑↑Stress
de Haro C et al Crit Care Med 2018;46:1385–92
Precise Information on DC is Lacking
In 7 hours
10620 10640 10660 10680 10700 10720 10740
0
1
2
3
4
nu
mb
er
of D
C
Power = 61 Duration (min) = 26
AUC = 30.5
2 min
Clusters of Double Cycling & Breath Stacking
* *** ** ** *
de Haro C et al. CCM 2018;46:1385-92
Male 39y. Moderate ARDS. 24 h MV. SAS 2-3. Day 1 (12.09pm)
Data Analytics in ICU
Sci Rep. 2018 Dec 4;8(1):17614
Sci Rep. 2018 Dec 4;8(1):17614
State transition for the Poisson hidden Markov models from the
time series indexed each 15 min.
Transition probability matrixState transition diagram
Male 39y. Moderate ARDS. 24 h MV. SAS 2-3. Day 1 (12.12pm)
Hospital / Critical Care
of the Future
Heterogeneity & Personalization
Data
Information Technologies & Prediction
Humans
Cross Disciplines = Cross Fertilization
Early Neurocognitive Rehabilitation in Intensive Care
Project funded by La Marató TV3 2010
Multidisciplinary Team
ENRIC at ClinicalTrials.Gov
JAMA October 24, 2018
Reduce job strain prevalence by improving the ability of ICU
nurses to cope with stressful situations:
Progress in the management of potentially iatrogenic situations
Analyze stress & emotions felt by nurses dealing with a difficult situation
Manage the quality of team relationships
Simulated scenarios
Debriefing sessions that discussed soft skills and practices
Conclusions: Among ICU nurses, an intervention that included education,
role-play, and debriefing resulted in a lower prevalence of job strain at 6
months compared with nurses who did not undergo this program
Humans & Artificial Intelligence
Artificial
Intelligence
(AI)
AI is the ability of computers to replicate
tasks that only humans could do in the past
Toll E. JAMA, June 20, 2012
physicianartist
family
Muchas Gracias !!