Clinical trials still rely on one-off clinic visits to measure health outcomes, but a new five-year fellowship aims to replace those snapshots with continuous data from wearables and phone apps. The problem is that digital health technologies—smartwatches, sensors, and apps—can track patients around the clock in their daily lives, yet regulators have approved only a single digital endpoint (for Duchenne muscular dystrophy). No standard statistical methods exist to validate these high-frequency measurements, analyse the resulting time-series data, or handle missing data when patients stop wearing devices. This fellowship will develop that missing methodology, using pulmonary hypertension as a test case, then extend the approach to heart transplantation and Parkinson’s disease. If successful, the work could transform how late-stage clinical trials are designed and analysed. Instead of asking patients to travel repeatedly to clinics for brief assessments, trials could capture meaningful health data from everyday life—reducing costs, burden, and dropout rates. The statistical framework and open-access code produced here would give trialists, clinicians, and regulators practical tools to adopt digital endpoints across many disease areas, quietly reshaping the infrastructure of clinical research.
View original technical description
Background Digital health technologies (DHTs), which include wearables, sensors and phone apps, have huge potential to change how data are collected in clinical trials, and consequently how trials are designed and analysed. Currently, endpoints (health outcomes) in clinical trials typically require in-clinic assessments and provide one-time snapshots of patients' health. DHTs can: Collect data in patient's daily lives (as opposed to in-clinic), which is more meaningful and relevant to patients; Collect high-frequency measurements, which can help answer novel clinical questions; Reduce costs and burden for patients by minimising the need to visit the clinic. However, transitioning from traditional endpoints, which are typically snapshot measurements, to digital endpoints, which involve high-frequency time series data, leads to open statistical questions in the design and analysis of trials. To date, there is limited statistical guidance on this transition and there is currently only one digital endpoint with regulatory qualification (for Duchenne Muscular Dystrophy). To enable a step change in the adoption of digital endpoints by trialists, clinicians and regulators, this fellowship develops statistical methodology to address following research gaps: Methodology for the validation of digital endpoints for late-phase trials, including decisions on the selection of the measurement period and choice of summary measure (Workpackage 1). Methodology for robust and efficient analysis of time series data from DHTs to better understand the treatment mechanism of action in early-phase trials, using Functional Data Analysis and Changepoint Detection (Workpackage 2); Methodology to appropriately handle intercurrent events and associated missing data in late-phase trials with digital endpoints (Workpackage 3). The fellowship aims to: Develop a methodological framework for digital endpoint validation, analysis of time series data and handling missing data to meaningfully embed digital endpoints in trials; Resolve these challenges for Pulmonary Hypertension as an exemplar and demonstrate the transfer of learnings to other disease areas, including Heart Transplantation and Parkinson's disease; Use insights from (1) and (2) to provide practical guidelines for designing and analysing clinical trials that use DHTs. Timelines for delivery The fellowship is planned for 5 years. See Gantt chart for the timeline of workpackages. Years 1 and 2 focus on methodological training, and Years 2 and 3 on cross-sector training. A placement at ICON is planned for Year 2 and I will organise a Knowledge Exchange event in Year 3. Years 4 and 5 focus on producing accessible tutorials, code and guidelines. Workpackage 2 will be supported by a PhD student and Workpackage 3 will be supported by a Research Associate. Anticipated impact and dissemination The fellowship aims to develop statistical methods for meaningful embedding of digital endpoints into early- and phase-phase trials through: Patient input on reasons for missing data and reporting of results from granular analyses; A Knowledge Exchange event to explore transfer of learnings to new disease areas with resources co-designed with a Patient Advisory Group to facilitate involvement from patient /public representatives; Publication of statistical methods, their applications and guidelines; Sharing open-access code and tutorials on GitHub; Dissemination through an implementation and adoption strategy.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know