Tracking Depression Dynamics in College Students Using Mobile Phone and Wearable Sensing

Area of Science:

• Digital mental health
• Computational psychiatry
• Mobile sensing

Background:

• Rising rates of depression among college students strain existing mental health services.
• Current mobile sensing approaches for mental health lack specificity to major depressive disorder (MDD) symptoms.
• A need exists for tailored digital phenotyping to accurately assess student mental well-being.

Purpose of the Study:

• To introduce novel 'symptom features' from passive sensing data to proxy DSM-5 defined depression symptoms in college students.
• To evaluate the efficacy of these symptom features in predicting depression among undergraduates.
• To assess depression dynamics throughout academic terms using validated screening tools.

Main Methods:

• Collected passive sensing data from smartphones and wearables of 83 undergraduate students over two 9-week terms.
• Developed and utilized custom 'symptom features' designed to reflect specific DSM-5 depression criteria.
• Administered weekly PHQ-4 and pre-post PHQ-8 surveys to track depression scores and dynamics.

Main Results:

• Identified significant associations between novel symptom features and self-reported depression scores (PHQ-8, PHQ-4).
• Demonstrated the capability of symptom features from mobile sensors to predict weekly depression status.
• Achieved 81.5% recall and 69.1% precision in weekly depression prediction.

Conclusions:

• Passive mobile sensing, using specifically designed symptom features, offers a promising method for continuous depression monitoring in college students.
• This approach can supplement traditional mental health services by providing objective, real-time insights into student well-being.
• The findings support the potential of digital phenotyping for early detection and intervention of depression in higher education settings.