RIACT: A Responsible AI System for Personalized Study Habit Tracking and Early Burnout Signal Detection in University Students

Burnout detection happens too late: The paper identifies that student burnout rates range from 12-70% in higher education—exceeding working populations—yet is "typically identified only retrospectively, after academic decline has already occurred," highlighting a critical gap in early intervention.

Transparency over black-box AI: RIACT uses "auditable rules rather than model judgement" for burnout warnings, meaning the system's alerts are based on explicit, understandable logic that can be examined and verified rather than opaque AI decision-making, making it easier to trust and validate.

Humble framing prevents harm: The system frames all outputs as "observations rather than a diagnosis," avoiding the risk of students over-relying on or being harmed by misclassified AI predictions about their mental health status.

Minimal data collection protects privacy: The application "limits data collection to self-logged behavioural fields" rather than collecting invasive personal information, reducing privacy risks while still enabling meaningful insights about study patterns.

Structured visibility enables self-awareness: A key finding is that "students have little structured visibility into their own study behaviour"—RIACT addresses this by having students log sessions and computing "net focus time by accounting for breaks," giving them concrete data to understand their own patterns and make informed decisions.