Intelligent detection of depression risk among college students based on campus data

Mental health issues among college students have become increasingly concerning, with depression being one of the most prevalent disorders. Traditional screening methods rely heavily on self-reporting questionnaires, which suffer from low participation rates and potential reporting bias. This study proposes an intelligent detection system that leverages multi-source campus data to forecast elevated Patient Health Questionnaire-9 (PHQ-9) risk at a scheduled future assessment. By integrating behavioral patterns from educational records, consumption data, and access control systems, we develop a machine learning-based early warning framework. Our approach combines feature engineering, temporal analysis, and privacy-preserving techniques to create a non-intrusive monitoring system. The contribution is a deployment-oriented integration of clinically grounded temporal features, leakage-controlled forecasting, privacy safeguards, and counselor-mediated review. Experimental validation on campus data demonstrates that the proposed method achieves promising forecasting performance while maintaining ethical standards and data privacy. The system provides a practical tool for universities to proactively support student mental health.