How to Harmonize and Interpret the Tsunami of Sensor Data in Clinical Trials

Author: Mark Matson, Managing Partner, Patient Cloud, Medidata

The surge of wearable sensors is transforming the clinical trial experience for patients, sponsors, and research sites. Sensors can be worn by the patient 24/7 to capture physiology continuously, providing new insights to a patient’s journey, measuring disease progression or effects from therapeutic intervention. The value of these data cannot be understated both as an asset to understanding disease processes and our ability to develop interventions.

One of the consequences of using high-frequency, high-fidelity sensors, despite all of the promise, is the amount of data collected, all of which requires analysis to identify physiological signals and, potentially, biomarker development associated with a disease or condition.

Data management and data science teams have a daunting task. They must convert raw data and computed metrics into meaningful clinical insights specific to the associated indication, often amounting to terabytes of data, which can quickly become a needle-in-the-haystack search for meaning.

How are sensor companies turning millions of data points into actionable insights to improve trials?

Sensor data have become a new source of clinical evidence in characterizing changes in patients based on therapeutic intervention. However, the increasing volume of data generated by sensors can quickly become overwhelming, making it difficult to store, manage, and analyze effectively, and so creating heretofore unknown challenges:

  • Data quality and compliance
    Devices selected for use in a trial must be fit-for-purpose to measure the physiology of interest. Once the device is selected, the collection protocol must be low-burden to support high compliance by study participants.Validation and reproducibility. The measure chosen must be validated in the target population (or a separate validation study should be conducted prior to or alongside the interventional study) to ensure that the clinical decisions taken and regulatory acceptance are optimized.
  • Lack of standardization
    Data collected and associated biomarkers are often measured using different technologies and protocols. This can lead to inconsistencies and challenges in data integration and interpretation.
  • Statistical power and significance
    High-frequency data and continuous biomarker collection will affect the number of participants required to achieve statistical significance. Understanding the interplay between the amount of data and the statistical power can save time and money.
  • Ethical concerns
    The use of continuous data and biomarkers in research requires a review to confirm the privacy, security, and consent procedures are adequate for this new type of data.

Medidata's Sensor Cloud solves the complex challenges of sensor usage in clinical research by ingesting vast amounts of sensor data and organizing it alongside other patient data to improve efficiency and effectiveness.

How can sensor data be used in trials?

Sensor data represent a wholly new form of clinical evidence to support endpoints in clinical research.  This data can be utilized in a number of ways:

  • Primary and Secondary Endpoints
    Traditional endpoints measure liquid or solid biopsy biomarkers or Clinical Outcome Assessments. Sensor data can augment those traditional measures by providing physiological information from the patient throughout the trial.
  • Real-time monitoring
    Sensors can enable real-time monitoring for critical adverse events to enhance patient safety and improve enrollment in complex trials.
  • Data analysis
    Overlaying sensor and traditional data allows researchers to surface insights automatically and construct machine learning models that advance our understanding of disease processes and therapeutic effects.
  • Security
    Using advanced encryption and access controls ensures that data is stored securely and can only be accessed by authorized users. This protects sensitive patient information and ensures compliance with regulatory requirements.
  • Collaboration
    Researchers want a platform that allows them to seamlessly share data, prepare analyses, test algorithms, and enhance collaboration across contributors to speed up the clinical trial process and improve the accuracy of the results.

How Sensor Cloud helps researchers manage sensor data

Medidata's Sensor Cloud provides a centralized location for ingesting, storing, and normalizing sensor data which allows researchers to access data in near real time. It also provides data analysis tools and automates insight development that can be used to inform clinical trial decisions.

The platform collects data from a range of sensors in trials including wearable devices, connected devices, and other medical-grade sensors. Once data is ingested, it is stored securely in the cloud, where it can be accessed by authorized researchers.

The scalability of the platform is designed to handle large amounts of data, making it suitable for clinical trials involving hundreds or even thousands of patients. Researchers can add new sensors to the platform as needed, without having to worry about the limitations of traditional data storage systems.  Data collected over time with various devices can be aggregated into singular datasets with the use of the common data model, enabling a future-proofing strategy previously unavailable to researchers.

Get the MDPI article discussing “A Reliable Methodology to Examine Long-Term Wearable Data Acquisition in Gait and Postural Sway,” a collaboration between the University of Vermont, Medidata Solutions (a Dassault Systèmes Company), and Stanford University.

The editorial staff had no role in this post's creation.