Prospective observational study on behavioral monitoring, disease progression assessment, and screening model development for patients with depression using wearable devices and mobile phones: protocol

Background – Depression affects over 95 million people in China (lifetime prevalence 6.8%), yet traditional assessments rely on episodic, retrospective evaluations missing dynamic symptom fluctuations. While digital phenotyping offers continuous monitoring potential, feasibility questions remain for psychiatric populations: patient retention, data completeness, and technical challenges. This study assesses the feasibility of intensive multimodal digital phenotyping in routine outpatient care and explores digital phenotypes in major depressive disorder (MDD). Methods and analysis – This is a prospective protocol for a 12-month observational cohort study at Peking University Sixth Hospital that recruited 202 MDD outpatients and 100 healthy adults (aged 18–60 years) from August 2023 to February 2025, with follow-up ongoing through February 2026. The study evaluates practicality and reliability of multimodal digital phenotyping in real-world clinical settings. Data collection integrates: (1) high-frequency clinician-administered assessments using the 17-item Hamilton Depression Rating Scale (HAMD-17) at seven time points for patients with MDD and at baseline for healthy participants; (2) daily ecological momentary assessments and biweekly self-report questionnaires; (3) passive smartphone sensor data (physical activity, GPS, screen usage, etc.); (4) continuous wearable data (sleep, heart rate variability, step count, etc.); and (5) environmental exposures (photoperiod, temperature, air pollution, etc.) linked to GPS coordinates. Primary outcomes include recruitment and retention rates, data completeness across modalities, and platform-specific constraints. Secondary analyses—prospectively planned pending full data collection—will (1) characterize digital phenotypes and temporal dynamics; (2) compare MDD vs. healthy participants; (3) explore environment-symptom associations. Results – MDD patients (n=202) were younger (30.8 ± 10.1 vs 42.8 ± 11.4 years) and more often unmarried (42.1% vs 15.0%) than healthy participants (n=100). Clinical-assessment completion declined from 100% (baseline) to 68.3% (Week 8). Among Android users (n=214), adequate EMA completion was 31.6% (MDD) vs 84.0% (healthy). Wearable adherence was 29.2% vs 70.0% (n=302). Conclusion – This feasibility study reveals marked compliance disparities between MDD patients and healthy participants, highlighting implementation barriers for real-world digital phenotyping in psychiatric populations.