Automated mood disorder symptoms monitoring from multivariate time-series sensory data: getting the full picture beyond a single number

Mood disorders (MDs) are a significant global health concern. This paper proposes a novel approach to MDs monitoring using wearable sensor data and machine learning (ML) to predict individual items of the HDRS and YMRS scales, rather than just overall scores. A deep learning pipeline achieved clinically significant agreement with expert clinician assessments (quadratic Cohen's κ and macro-average F1 score of 0.609). The study addresses challenges like multi-task learning, class imbalance, and subject-invariant representation learning, highlighting the importance of out-of-distribution testing.

Filippo Corponi, Bryan M Li, Gerard Anmella, Ariadna Mas, Isabella Pacchiarotti, Marc Valentí, Iria Grande, Antoni Benabarre, Marina Garriga, Eduard Vieta, Stephen M Lawrie, Heather C Whalley, Diego Hidalgo-Mazzei, Antonio Vergari