Unlock the Value of Your Clinical Trial Data and Content ... critical component of clinical trials. To truly enable Data Science in clinical development, these tools need to draw from

05/2017 WHITE PAPER

Medidata and other marks used herein are trademarks of Medidata Solutions, Inc. All other trademarks are the property of their respective owners.Copyright © 2017 Medidata Solutions, Inc.

Unlock the Value of Your Clinical Trial Data and Content with Big Data Discovery

205/2017 WHITE PAPER UNLOCK THE VALUE OF YOUR CLINICAL TRIAL DATA AND CONTENT WITH BIG DATA DISCOVERY

Big Data has Redefined Our World

From leisure activities, to our own health and wellness, to the industries in which we work — Big Data has transformed our world. Subscription-based content providers, like Netflix and Amazon Prime, are changing television programming by using detailed customer segmentation and viewing habits to rethink how new programming is funded, produced, and released to the market. Everyday items like Nest are transforming home heating and cooling by collecting and aggregating sensor data to automate thermostat changes.

In the clinical research industry, the creation and availability of many more data points is changing everything too. Historically, data came from a single source: a patient visiting a clinic, whose information would be entered into an electronic data capture (EDC) system. Today, clinical trials are accommodating an incredible variety of data and content sources: from traditional clinical data, to high resolution images, to genomic and wearable sensor data, investigator files, consent forms, and much more. This data explosion brings new and transformative opportunities, but it also comes with additional risks.

Big Data is Not the Entire Picture

According to Gartner, Big Data is only one element of a three-part model: Big Data, Data Discovery and Data Science combine to create a model that Gartner defines as Big Data Discovery. In clinical trials, Big Data Discovery centers on creating and using interactive reports along with exploring data from multiple sources, such as imaging and mHealth data, to develop actionable insight during clinical development. Ultimately, these three elements of the model require simpler tools, for a wide range of users, and access to a standardized and broad variety of data sources.

Source: ZDNet.com & Gartner. Big Data Discovery combines Big Data, Data Science, and Data Discovery


1. Big Data Today the technology industry has settled on five key characteristics when describing Big Data: Variety, Volume, Velocity, Veracity and Value. Clinical trials now collect data from a variety of sources that go beyond traditional in-clinic data. Each source presents a unique opportunity to supplement the effectiveness of a clinical study, while they also present unique challenges. For instance, medical images, X-Rays, CT Scans, and MRIs offer greater insights into the overall patient profile, but they are often stored and managed separately from systems capturing traditional clinical data.

Second, clinical researchers must consider the sheer volume of Big Data. Many of the newly identified forms of data are inherently large. A single MRI can be multiple gigabytes. Creating an IT architecture that allows practitioners to upload, download, modify, and query a multitude of data can be a challenge.

Velocity relates to how frequently data is collected. For example, sensor data provides researchers with multiple measurements per patient, per second.

Data variety, volume and velocity inevitably lead us to veracity: how can we ensure accuracy and integrity in our clinical trials? Essentially, data collected from various sources must be cleansed, standardized and verified.

Finally, if leveraged correctly, combining the variety, volume, velocity, and veracity of data creates a robust and standardized data set that can power the analysis and insight that can ultimately unlock tremendous value when conducting a clinical trial. Data and content can be leveraged by clinical researchers, sponsors and CROs alike to optimize every aspect of the process. Collecting and standardizing Big Data on its own without instituting Data Discovery and Data Science is not sufficient to unlock insights that will ultimately drive value across the clinical development process.

2. Data DiscoveryData Discovery tools like dashboards and benchmarks provide ease, agility, and flexibility in data analysis that can be accessed by a wide range of stakeholders who aren’t required to have specialized skills in data science or statistical modeling. Many have termed these new stakeholders as Citizen Data Scientists.

In clinical trials, data discovery is paramount. It’s not necessarily about analyzing all of the data - but rather surfacing the relevant data to support decision-making around a specific use case like patient engagement, site feasibility or site monitoring. Ensuring cross-study reporting and ad hoc reports that target specific study parameters can enable a multitude of small study adjustments, thus having a large impact on feasibility or performance.

To unlock the value of Big Data, clinical development technology must go beyond being able to ingest, standardize and verify large volumes, varieties, and velocities of data to providing reporting and dashboarding capabilities that are easy to consume. Doing so enables the vast ecosystem of those working on clinical trials — whether in data, monitoring, operations, or regulatory domains — to optimize the various processes that underpin their daily activities.

3. Data Science While Data Discovery can empower Citizen Data Science across the clinical development ecosystem, Data Science and the analytical capabilities it can power offers several opportunities to transform clinical development. These include advancements in Site Selection, Risk Based Monitoring, and even Synthetic Control Arms. More importantly, the extent to which data science can enhance clinical trials is directly proportional to the robustness of its data set. An analogue to this concept is the wisdom of the crowds. When the life science industry arrives at solutions that combine and standardizes data across tens of thousands of trials, therapeutic areas, geographies, and sponsors, it enables a richer dataset.


Combining these vast datasets with new Data Science capabilities like artificial intelligence and machine learning can shed light on unknown relationships between clinical trial factors. In clinical trial operations, the application of Data Science tools not only supports the Data Discovery process in terms of dashboards and benchmarks, but it can also directly impact the trial in recruitment, site management, adverse event prediction and drug supply.

For example, machine learning algorithms can learn from historical clinical trial data to automate the standardization of verbatim adverse event descriptions. This standardization may allow for improved signal detection in a trial, so that safety risks may be identified sooner. Similarly, algorithms can be trained to detect whether such an event is serious by looking at its associated clinical features. This automated classification can reduce the reporting time to regulatory agencies, which is a critical component of clinical trials. To truly enable Data Science in clinical development, these tools need to draw from a flexible platform that ingests, standardizes, and aggregates data, while adding intelligence without impacting data fidelity.

Unlocking the Value of Big Data Discovery in Clinical Trial Operations

The EDC holds the key to Big Data but it won’t unlock Data Discovery and Data Science on its own. As data is ingested from different sources, a powerful way to aggregate it is through an Electronic Data Capture (EDC) system. The most advanced EDCs in clinical development have evolved beyond simply capturing traditional in-clinic data to ingesting data from multiple sources, while simultaneously cleaning, standardizing and verifying it.

At its core, an EDC provides a single data store — all in one place — for all collected data, thus becoming a study’s single source of truth. Its singular dataset can then be leveraged to drive the Data Discovery and Data Science to report, analyze and make study conduct decisions in real time.

Ultimately, the EDC can only set the stage. Next comes putting the Data Discovery and Data Science technology and human capital needed to deliver the capabilities into the hands of the clinical development practitioners who can leverage these insights to take action. This is often a critical decision point for a clinical development organization and its key partners in IT.

In addition to focusing on effectively managing the pipeline of studies currently being conducted, do these departments want to build the internal technological and data scientist capabilities to support Data Science and Data Discovery? Or should they leverage clinical development technology providers who have built Data Discovery and Data Science capabilities into their solutions?

Today the most advanced clinical development technology providers offer Data Discovery and Data Science capabilities as a service. Taking advantage of these capabilities can lessen the future development burden on clinical development and the IT organizations that support them. As the rest of this paper will explore, leveraging these capabilities can also power the trials of the future, including intelligent risk-based monitoring and adaptive trials.

Below are three use cases that illustrate how the principles of Big Data, Data Discovery and Data Science can be applied during the course of a clinical trial.


Leveraging Big Data for Adaptive Trials

The ability to review data as defined in the adaptive trial protocol, and implement changes to the study design in real time can enable new breakthroughs in adaptive trials. Live study design changes allow sponsors and partners to scale up and down more effectively and efficiently within minutes. With these types of capabilities, very soon all sponsors will be able to easily execute highly complex trial designs.

What’s the promise of adaptive trials? Adaptive trials are those with a prospectively planned opportunity for modification of one or more aspects of the study design and hypotheses, based on analysis of interim study data from subjects. Analyses of the accumulating study data are performed at prospectively planned timepoints within the study in a fully blinded or unblinded manner, and can occur with or without formal statistical hypothesis testing.

Compared to non-adaptive studies, adaptive design approaches may lead to a study that more efficiently provides the same information, increases the likelihood of success on the study objective, or yields improved understanding of the treatment’s effect (e.g., better estimates of the dose-response relationship or subgroup effects, which may also lead to more efficient subsequent studies). Further, adaptive trial design enables the ability to titrate based on the dose response for patient safety, which is particularly important with high-risk disease areas.

What are the problems? Upfront planning and strategic thinking are required to ensure adaptive trials are successful. Combining trial designs in one study can drive deeper analysis. Yet this requires extensive planning and a considerable investment in setup time, adding weeks whenever a change is made. In some cases, adaptive trials are disregarded altogether because of their complexities. Because adaptive trials require so much planning, a sponsor may lack the time or resources to plan the design adequately. Adaptive trials are most often seen in oncology or rare diseases.

One solution is a randomization and trial supply management (RTSM) technology with a live study design feature, allowing users to implement study design changes within minutes. Sponsors and CROs can add, disable and change treatment compositions, create new dosing rules based on new treatment compositions and change current patient treatments.


Data’s role in adaptive trials The two aspects of a trial with the greatest potential for rapid change are the way randomization is performed and the way controls are selected. Adaptive trials are an example of both: in one such trial, breast cancer patients are randomized into a particular arm of the study based on biomarkers, but they can move from one group to another as more information becomes available. Thanks to Big Data, there is also a shared placebo arm that, over time, can be cross-referenced with other studies. In adaptive study design, Phase I-II trials can elapse much more quickly and Phase III trials can be shorter. What results is a cut in costs and time, and less patient exposure to experimental molecules.

Medidata OPAL Diagnostics Uses Data Discovery through Interactive Benchmarks Against Industry Data


Leveraging Data Discovery Concepts for Regulatory and Nonregulatory Content

Clinical trials don’t just generate data. They also generate a wide array of both regulated and nonregulated content. Cloud content platform can enable real-time collaboration, faster access to archived content and easier ways to search and discover related content.

What’s the promise? From trial master files, contracts, CVs, IRB letters, SOP workflows, and more — there’s a variety of content from multiple sources in clinical trials. There’s also an increasing flow of content, with online content platforms enabling multiple users to simultaneously create, edit and approve various pieces of content in real time. Various cloud-based technology platforms, embracing Data Discovery, help these users create a ubiquitous, streamlined and interactive network, that not only improves productivity but increases regulatory compliance and decision making.

What are the problems?Content is often spread across silos, which changes how organizations manage and regulate content. This can prove difficult during regulatory inspections, when the ability to deliver auditor-requested content quickly and accurately is an imperative. It’s critical that regulated content and its metadata are 21 CFR Part 11 compliant with the accompanying workflows.

With sponsors and CROs needing multiple platforms to manage both regulated and nonregulated content, implementing and maintaining these systems is costly. Non-intuitive systems also lead to significant user frustration and process compliance issues, which can result in regulatory body compliance issues with serious implications. How can you manage many different sources of data and information to ensure collaboration, inspection readiness and compliance?

The Solution in Regulated Content ManagementHaving an integrated, end-to-end system for regulated content management (RCM) — delivering required processes and ensuring confidentiality, high integrity, traceability and availability — is essential for success. Hayley Lewis, VP of Regulatory Affairs and Quality at Zosano Pharmaceuticals, discusses the importance of employing a quality RCM solution in the clinical trial space:

“Within this industry almost everything is regulated, so having traceability and an applied process of doing things is extremely important. Having your documentation in a secure, validated environment allows ease of use for authoring, review and approval; this also enables you to focus attention on other key areas of importance in the trial.”


Leveraging Data Science in Clinical OperationsAs clinical patient data is collected during a study, there is also related metadata that has value: therapeutic area, site location, payment status and various process durations, when aggregated across hundreds or thousands of other studies, can provide a powerful dataset to enable analytics.

What’s the promise of metadata? Predictive analytics in clinical operations can help optimize some of the most important processes and benchmarks within a trial. Using historical site performance to select sites for a future clinical trial can dramatically increase patient recruitment and study start time. Using machine learning anomaly detection can automatically flag data that needs further investigation, leading to risk-based monitoring. Leveraging these types of predictive analytics in clinical operations can shorten trial timelines, improve data collection accuracy, and ultimately improve efficiency.

What are the problems? For one, successfully capturing metadata using predictive analytics requires a fairly vast dataset and a large repository of historically similar trials. Additionally, this dataset must be standardized and anonymized for patient privacy and protection. Data science expertise is also required to conduct the appropriate statistical modelling, thus delivering the key insights needed for action.

Clinical trial technologists are solving these problems by leveraging software and solutions that enable the collection, aggregation, and analysis of clinical trial metadata. Using a strategic monitoring solution with intelligent, risk-based monitoring, sponsors can perform a comprehensive scan of clinical trial databases for inconsistencies across data domains, sites and patients. Once anomalies are detected, site monitors can directly target those anomalies — instead of costly and time-consuming 100% SDV.

Medidata Centralized Statistical Analysis Uses Machine Learning to Power RBM


In Conclusion

The creation and availability of more data points is changing everything — from our work and leisure activities, to our health and wellness. Clinical trials are accommodating everything from traditional clinical data, to high resolution images, laboratory testing, regulatory files, genomic and mHealth data. Truly understanding the benefits and challenges of the three technologies — Big Data, Data Discovery and Data Science — allows us to bring our trials to life and gain deeper, more meaningful insights and real value from areas like adaptive trials, clinical operations, and regulated content management. These hold the key to more efficient trials.

About MedidataMedidata is reinventing global drug and medical device development by creating the industry’s leading cloud-based solutions for clinical research. Through our advanced applications and intelligent data analytics, Medidata helps advance the scientific goals of life sciences customers worldwide, including nearly 850 global pharmaceutical companies, biotech, diagnostic and device firms, leading academic medical centers, and contract research organizations.

The Medidata Clinical Cloud® brings a new level of quality and efficiency to clinical trials that empower our customers to make more informed decisions earlier and faster. Our unparalleled clinical trial data assets provide deep insights that pave the way for future growth. The Medidata Clinical Cloud is the primary technology solution powering clinical trials for 17 of the world’s top 25 global pharmaceutical companies and is used by 16 of the top 20 medical device developers—from study design and planning through execution, management and reporting.

[email protected] | mdsol.com | +1 866 515 6044

Medidata Clinical Cloud®

Cloud-based clinical research solutions | Innovative technology | Data-driven analytics Reduced costs | Improved time to market | Faster decisions | Minimized risk

Medidata OPAL Site Selection Uses Benchmarking of Historical Site Performance to Aid Site Selection

Documents

Unlock the Value of Your Clinical Trial Data and Content ... critical component of clinical trials. To truly enable Data Science in clinical development, these tools need to draw from