ABSTRACT: This research proposes an explainable artificial intelligence (AI)-based framework for monitoring student academic development and supporting early identification of academic risk among school students. The proposed system is intended for approximately Standards 1–10 and integrates academic performance, attendance, assignment completion, behavioural and engagement observations, student health-related information, and selected family-environment and socio-educational factors. These factors may include recurring health difficulties, family circumstances, parental relationship context, and emotional support that may be associated with student well-being and academic development. Machine learning models will be evaluated for academic performance prediction and academic-risk classification, while explainable AI techniques will be used to communicate the factors influencing individual predictions. The framework is designed as a decision-support and early-support system rather than a diagnostic or autonomous decisionmaking tool. When a decline in performance is detected, the system will present model-identified factors and supportive recommendations for review by authorized teachers and parents or guardians. Particular attention is given to privacy, fairness, data minimization, security, human oversight, and the protection of children. Actual experimental results will be reported only after implementation and evaluation; therefore, no performance values are claimed in this paper.
KEYWORDS: Student Performance Prediction; Educational Data Mining; Machine Learning; Explainable AI; Academic Risk; Student Development; Early Intervention; Student Health; Family Environment; Student Wellbeing; Educational Support.
AN EXPLAINABLE AI-BASED FRAMEWORK FOR EARLY IDENTIFICATION AND SUPPORT OF STUDENT ACADEMIC AND DEVELOPMENTAL RISK
MRS. ANJALI DANDEKAR, MR. RAJ KADAM