Assessing Depression Severity Through a Passive Digital Mood Marker

ObjectiveEffective management of Major Depressive Disorder (MDD) is limited by reliance on episodic, subjective clinical scales. Passive digital phenotyping offers a potential solution for continuous, objective monitoring. We aimed to assess the concurrent validity of a novel digital biomarker--the Facial Affect Dynamics-derived Depression Severity (FADS) score--against the Patient Health Questionnaire-9 (PHQ-9). MethodsWe conducted an interim analysis of the EMC2FR study (NCT06860165), a prospective, multicenter cohort study in France. Seventeen outpatients with unipolar MDD used the EmoDTx smartphone application, which passively analyzes facial expressions via the front-facing camera during routine device use. We analyzed data from Week 2 (W2) and Week 6 (W6) visits. The association between the 14-day passive FADS score and the PHQ-9 was estimated using Spearman rank correlation with participant-level bootstrapping to account for repeated measures. ResultsThe analysis included 34 person-visit observations from 17 participants (70.6% women; mean age 42.7 years). We observed a strong, positive concurrent correlation between FADS and PHQ-9 scores ({rho} = 0.65; 95% CI, 0.50-0.87; P =.01). Furthermore, within-participant changes in FADS scores from W2 to W6 were strongly correlated with changes in PHQ-9 scores ({rho} = 0.67, P =.02). ConclusionsPreliminary results suggest that passive monitoring of facial affect dynamics is associated with clinically relevant variation in depression severity. These findings support the ongoing EMC2FR study to validate this marker as a tool for longitudinal depression monitoring.