To propose a behavior-first framework for explainable mental health early warning in higher education that addresses implementation and governance challenges related to digital health systems.
Approach:
Generative Semantic Intermediary Framework (GSIF): GSIF is organized around three layers: ecologically feasible multimodal observation, two-stage generative semantic translation, and constrained review prioritization.
Key Findings:
Current mental health monitoring methods in universities face challenges such as intrusive data collection and limited auditability.
Digital phenotyping and predictive AI can characterize changing risk trajectories but raise ethical and practical concerns.
The GSIF framework emphasizes data minimization, human-in-the-loop verification, and explicit escalation thresholds.
Interpretation:
The GSIF framework aims to enhance the transparency of the pathway from routine data to review recommendations.
Limitations:
The feasibility and acceptability of the GSIF framework have yet to be evaluated.
Potential false-positive burdens and reviewer calibration issues need further investigation.
Conclusion:
Future work should evaluate the GSIF framework's effectiveness compared to existing screening and monitoring approaches.
