By Kara Dennis
Health care providers and government agencies are looking for ways to better understand the patient experience, disease progression and therapeutic response, and companies across the health care value chain are building capabilities to manage and analyze large-scale data sets for these purposes.
Ultimately, the winners in this space will develop analytical tools and approaches that enable health care practitioners to become smarter and more responsive with each new patient treated. That is, as each patient receives a treatment, the data gathered on that patient and his or her response will help clinicians better understand how to treat their next patient by using Big Data technologies for a faster and more informed bench-to-bedside feedback loop.
Let’s explore how Big-Data-driven technologies might play out in clinical research. Traditionally, subjects in clinical trials participate in clinic-based visits on a periodic basis, providing investigators with snapshots in time of disease progression or therapeutic response. The data gathered in these clinic visits is tremendously valuable; however, the clinical trial sponsors I speak with every day tell me that they often find existing endpoints to be insufficient—at times describing them as “subjective,” “not sensitive or nuanced enough,” or “incomplete.”
Today, we can gather much richer data by equipping patients with sensors, wearable devices, and mobile apps. By combining biometric and activity data with traditional site-based data, genomic information, and patient-reported outcom es, research teams have access to unprecedented information about the patient experience and are able to move closer toward making personalized medicine a reality.
Validating and Evaluating Big Data
Once real-world data is gathered, the first application of Big Data is to evaluate its quality, determining whether it is comprehensive, accurate, and attributable.
By analyzing this data in context with other clinical and genomic patient data, the industry may find answers to wide-ranging clinical questions, since the combination of traditional clinical data, patient-reported outcomes, and sensor-derived cognitive behavioral phenotypic data provides a more complete picture of the patient. This is where Big Data becomes critical—high-resolution sensor data can be vast and often requires best-in-class tools for scalable data collection, analysis, and storage.
For example, an area of great interest for sponsors right now is patient movement, which has the potential to be used as a proxy for quality of life. Another emerging area of interest is cognitive tasks. Mobile apps are enabling researchers to gain insight into neurological disease severity based on patient interaction with or reaction to mobile tests, games, and/or audio-visual stimuli..
Putting the Patient at the Center
At Medidata, we spend a lot of time working with sensor manufacturers to make sure they are developing tools that can be effectively used in clinical trials. We’ve run a number of these studies with sponsors, and we’ve realized that as the patient experience improves, so does compliance. If you provide a wearable device that requires minimal effort from the patient—such as a long battery life and simple instructions—engagement is higher and so is the quality of the data.
Across the life sciences industry, we have seen considerable investment in mHealth technology, often in support of therapies for the management of chronic disease. We’re working with many sponsors right now that are interested in approaches that have the potential to decrease site visits or procedures during a trial. These sponsors are adopting devices that enable patients to take measurements, like their blood pressure, at home rather than visiting the office. Many of these devices integrate with hubs that can passively sync data from sensors and devices to the cloud without any patient intervention; this ensures that all data is captured while also lowering the burden for patients.
Regulatory agencies like the FDA have made strides to improve patient engagement across clinical research. The 2013 Patient-Focused Drug Development Initiative reinforced the “importance of developing a systematic method of obtaining patients’ point of view.” Chronic, symptomatic diseases that affect daily life are a key focus area for the agency.
With that as a starting point, clinical trial sponsors are very interested in better understanding their patients, their disease, and their response to therapy. Increasingly, many sponsors are seeing this as a competitive advantage—to understand their patients better than anyone else.
Operational Implications
I would be remiss if I didn’t mention the operational aspects that must be considered. Both the infrastructure to support Big Data and the investment in sensors exist, but there’s also a really critical operational piece, which we refer to as “the last mile.”
When mobile devices or sensors are used in a clinical trial, investigative sites must be properly prepared to equip patients with devices and explain how to use them. In turn, patients need to understand their responsibilities, such as when and how to charge devices or connect to the Internet. In the same vein, clinical trials must have a plan in place for lost or broken devices so that the trial can continue seamlessly. We provide visualization tools so that sponsors and investigative sites can monitor this operational management, and we are investing heavily in learning resources and support services for these trials.
The end result of all of this is to evaluate therapeutic impact. For this, much more interesting exploratory analysis is needed, one that requires deep understanding of the disease and the wearable that is being used, and a broad toolkit of data science techniques (e.g., digital signal processing and machine learning) to find signals.
The clinical research landscape is evolving with the availability of high-frequency, high-resolution data on a variety of behavioral and cognitive phenotypic elements, and it’s evolving with the improved infrastructure and tools to process this data. When you combine these elements, the result can be powerful and has the ability to change how we measure endpoints in clinical trials.
Kara Dennis is managing director of mHealth at Medidata Solutions.
