Digital Phenotypes for Predicting Depression

This longitudinal study aims to identify and validate digital phenotypes that can predict recurrence of major depressive episodes using passively collected, real-time sensing data from smartphones and wearable devices. Over a 12-month period, 540 participants-including patients with mood disorders and healthy or high-risk controls-will complete five clinical assessments at 3-month intervals, wear a Fitbit device daily, and log daily mood ratings via a mobile app. The study includes the development of AI-based predictive models and the construction of an anonymized wearable big-data repository for mood disorders.