Digital Phenotyping of Isolation and Loneliness (DPhIL)

We will utilize emerging methods to integrate temporally dense smartphone-based data to better characterize human social engagement (e.g., conversation detection, geographic location, digital communication), with the ultimate aim of reducing social isolation. Smartphone sensors provide continuous, unobtrusive, and real-time assessment that can be used to predict momentary transitions to isolation and loneliness. We will develop and test algorithms that incorporate both active (ecological momentary assessment) and passive (movement, location, conversation) metrics to improve prediction of social isolation and loneliness. We will incorporate methods that optimize characterization of social behavior from smartphone sensors (Aim 1), evaluate the promise of a network-based approach to predicting transition states of social isolation and loneliness in a healthy population (Aim 2), and apply our network approach to a clinical sample characterized by social isolation (Aim 3).