Automated Mood Disorder Symptoms Monitoring From Multivariate Time-series Sensory Data: Getting The Full Picture Beyond A Single Number

This article discusses the use of wearable technology and machine learning to monitor and predict mood disorders, specifically focusing on the use of physiological data to infer symptoms and assess treatment needs. The authors propose a new task of inferring all items in the HDRS and YMRS scales, and develop a deep learning pipeline to score symptoms with high accuracy. They also address challenges such as class imbalance and subject-invariant representations.