AI-BASED EARLY DETECTION OF ACADEMIC STRESS AND BURNOUT AMONG UNIVERSITY STUDENTS: A REVIEW

Academic stress is an important issue in higher education and can occur as a result of the workload, examinations, deadlines, competition, financial problems, uncertainty, and adjustment to university life. If stress persists, it can lead to exhaustion, disengagement, decreasing effectiveness in studies, and burnout. The research proposal to which this review is based on states the necessity to develop scalable and ethical ways of detecting students with academic stress who risk moving to serious distress or burnout. In this research paper, AI and ML are proposed to be used for early detection of the risk. The review analyzes the recent evidence on academic stress and the burnout of students based on AI/ML detection and prevention technologies with special attention to the data sources, ML methods, multimodality, wearables, explanations and validation, privacy, ethical issues, and human oversight. The provided literature map shows the most common limitations including cross-sectional design, survey dependence, small or one-discipline samples, limited external validation, as well as the lack of reasoned connection between prediction and intervention. The study suggests that future investigations ought to progress from static categorization towards longitudinal forecasting through employment of academic, behavioral, psychological, digital-engagement, and, where ethical considerations permit, biological input. A responsible early warning system is expected to provide valid results, with temporal and external validation, internal calibration, subgroup fairness assessment, interpretable output, as well as ensuring privacy of personal information being processed. The paper wraps up the investigations with the provision of a concept research framework suitable for the higher education context and in line with the proposed contribution to students’ welfare and quality education.