In response to the practical constraints of “large sample size, limited manual review, and insufficient discrimination of single total scores” in the normalization of psychological screening in universities, this article studies the identification and feature analysis technology of psychological health risks among college students. Based on publicly available psychological assessment data from 24292 students, integrating PHQ-9, GAD-7, PSS, ISI scales, demographic information, and question by question response time, a student level risk representation was constructed. A multi view recognition framework with enhanced response time was designed to incorporate symptom intensity, response time deviation, and similar context of students into the unified discrimination process. The results showed that in the publicly available samples, PHQ-9 mil and above accounted for 30.49%, GAD-7 mil and above accounted for 16.97%, PSS high and above accounted for 11.53%, and ISI threshold and above accounted for 9.06%; The coverage of stress-related burdens is wider, and the marginal zone of depression risk is longer. Further analysis of response behavior shows that there is a clear bimodal structure in the total response time of ISI and PSS, with key turning points located at approximately 12 s and 23 s, indicating that response time can provide incremental information for edge risk identification. This article presents a low-cost risk identification and interpretation analysis solution for university scenarios, which can provide technical support for screening stratification, manual review sorting, and referral decision-making.
