P.4.004 - Passive behavioural monitoring in neuropsychiatric disorders using smartphone technology

Abstract

The majority of medical disciplines utilize a measurement-based approach for the prevention and early detection of major medical events by monitoring measures that indicate the onset of such an event. However, in psychiatry treatment is often preceded by a major psychiatric crisis or even hospitalization and is followed by a long period of disability and slow recovery. The focus on acute care in psychiatry is due to a lack of valid temporal and objective measures of behavior that translate into day-to-day functioning. Several Behavioral changes in day-to-day functioning are identified as prodromal symptoms for several neuropsychiatric disorders [1,2]. A measurement-based approach is needed in psychiatry to monitor daily functioning and improve patient outcome [3]. Therefore, we have developed a passive behavioral monitoring (PBM) application, called BeHapp (https://behapp.org/), that collects a rich longitudinal trace of real-world social and behavioral data with minimal awareness. PBM is based on the ubiquity of smartphones, and monitors behavior by utilizing the large extend of smartphone sensor modalities available. These modalities allows us to formulate several features related to social behavior, e.g. number of contacts, duration of phone calls, social media usage, and places visited. We studied the predictive power of PBM in differentiating between non-diagnosed controls and neuropsychiatric patients by using smartphone modalities that relate to social behavior. We collected preliminary data from 24 neuropsychiatric patients (9 Major depression, 11 Schizophrenia, 4 Alzheimer’s Disease) and 72 non-diagnosed controls by installing BeHapp on the participant’s own android smartphone. Participants were included over several different ongoing studies and psychiatric screening data was available for 20 non-diagnosed controls. Average participation time was 13.4 days (sd = 5.9), variation is due to early removal of the application. We used GPS, application usage, and WiFi data to generate features that translate into daily social functioning. Simple two-sided t-tests were used to evaluate the difference in means between the neuropsychiatric patient and non-diagnosed control group. Additionally, we applied t-distributed stochastic neighbour embedding clustering [4] to group participants into a two-dimensional space regardless of their diagnosis. Our preliminary results revealed social-behavioral differences between the two groups (pooled patients versus controls) in social media usage; frequency of WhatsApp usage (p Our results suggest that PBM is sensitive in detecting differences in day-to-day social functioning and therefore a potential candidate for a measurement-based approach in psychiatry.