Key points are not available for this paper at this time.
Given the prevalence of depression among young adults, particularly those aged 18-25, this study aims to address a critical need in higher education institutions for proactive, private, automated mental health self-awareness. This study protocol outlines how a mobile phone application will leverage sensor signal and survey data to develop an automated screening tool for depressive behaviors. By analyzing sensor-based behavioral data through deep learning techniques, the proposed study seeks to identify students exhibiting depressive symptoms and their specific behaviors. Approximately 1,000 first-year undergraduate students (age 18 and above) will be recruited from two public US universities, one in the Midwest and one in the Southwest. For the midwestern university, there will be 11 surveys (baseline, nine follow-ups, and an endline) collected throughout a single academic year (2024-2025). However, at the southwestern university, only nine surveys will be administered during a semester. Simultaneously, sensor-based behavioral data on behaviors such as physical activity, social interactions, and sleep will be continuously collected passively. The main analysis will focus on understanding the relationships between human behaviors captured by sensor-based behavioral data and self-reported mental health surveys. Machine learning and deep learning algorithms will be used to uncover key behavioral patterns most indicative of mental disorders such as depression.
Colato et al. (Thu,) studied this question.