PocketPPD smartphone sensing screens postpartum depression with 83% accuracy

Postpartum depression (PPD) affects about 20% of new mothers, but traditional screening relies on burdensome self-report questionnaires or active logging, limiting long-term use. In a new paper on arXiv (July 2026), researchers from the University of Tokyo, National Center for Global Health and Medicine, and the University of Washington introduce PocketPPD, a system that uses passive smartphone sensing—collecting data on behavioral rhythms, stability shifts, and routine volatility—to detect PPD risk without requiring any active input from users. The team conducted a 4-week feasibility study with 61 postpartum women, using off-the-shelf smartphone sensors to capture multimodal data.

PocketPPD’s passive-only model achieved an AUC of 0.75, while the best-performing model that integrated both passive and self-report features reached an AUC of 0.83. Notably, morning and late-night routine volatility emerged as the top digital biomarker, with its predictive power dynamically moderated by maternal context such as the infant’s developmental stage and the mother’s employment status. This work provides empirical evidence that passive sensing can accurately screen for PPD, paving the way for low-burden, continuous perinatal mental health monitoring using devices already in mothers’ pockets.