Mobility-Based Community Analysis for Early Detection of Complex Psychiatric Disorders

Diagnosing psychiatric disorders, such as ADHD and anxiety, presents multifaceted challenges. While patients often exhibit multiple disorders, data collection often suffers from a reliance on self-reporting and subjective mechanisms. These challenges complicate accurate diagnosis and make it different to chart effective personalized treatments. It has been reported by various studies indicate that altered mobility may serve as a signal or a biomarker for identifying such disorders. However, much work remains needed to design well-defined and robust approaches to achieve such an objective. In this study, we propose a community-based network analysis approach that is based on utilizing mobility data collected from wearable devices. The first step in the proposed approach is to develop a graph model from the mobility data. Then, a network-based community detection algorithm is applied to identify inherent groupings of similar mobility characteristics within the population of subjects. The acquired communities are then comprehensively characterized through an integrated enrichment analysis involving clinical features, severity scores, and mobility patterns. The obtained results show that individuals within each community manifest a spectrum of psychiatric disorders, shedding light on the intricate interplay of these conditions. Furthermore, the study reveals distinctive patterns in sleep-wake cycles, daily movement behaviors, and the intensity of movement acceleration within these communities. These patterns not only serve as potential differentiating signals or early biomarkers but also offer a new understanding of features associated with psychiatric disorders for identification and characterization purposes. By integrating mobility data into the diagnostic framework, the introduced approach enhances the precision and relevance of psychiatric interventions, fostering a new paradigm that adapts to the unique needs of each patient.