Drawing on collaborations with Stanford University through the My Heart Counts app, Allan will discuss the potential of combining continuous wearable data with healthcare records. He will share key lessons learned from model development and validation, including why promising models may fail to generalise across different populations and how differences in control groups can influence performance.

Allan will also present current work being conducted in parallel in the United States and the United Kingdom, including the development of a cross-platform digital health research infrastructure that integrates data from Android and iOS devices, as well as wearable technologies such as Oura, Fitbit, and Withings. Through the inclusion of ECG data, raw sensor data and general PPG, this research aims to develop some generalizable, translatable models not just for early detection but also remote monitoring.

Finally, this session will also explore the next phase of the platform: using large language models and behavioural science to deliver personalised, adaptive health interventions. By combining prompt engineering, reinforcement learning, and real-time behavioural feedback, the project aims to optimise “just-in-time” digital nudges that encourage healthier behaviours based on an individual’s characteristics and responses.