Original title: AI-Driven Early Identification of At-Risk Students Through Learning Management System Behavioural Analytics
Lawal G. Anand · August 2026
Schools often notice a student is struggling only after they fail an exam, when it is already hard to fix the problem. To solve this, researchers designed a system that spots warning signs early by looking at day-to-day school software habits. Learning Management Systems are online platforms where students turn in homework and read class materials. Instead of waiting for test scores, the proposed system tracks daily actions. It checks how often students log in, how long they read materials, when they submit assignments, how they attempt assessments, their discussion posts, and gaps where they do nothing. Computer programs analyze these actions to separate engaged students from those likely to fall behind or drop out. This gives teachers and support staff early alerts to help students before grades drop. The study notes that schools must balance this tool with human judgment, privacy, and fairness.
Imagine a coach watching players in practice, not just at the final match. The coach notices one player starts arriving late, putting in less effort, and missing drills. By the time the actual match arrives, the decline is obvious — but by then it is hard to fix. In this paper, the online learning platform is the practice field, and the students' login patterns and assignment behaviour are the drills. The system watches those everyday actions continuously so it can flag a player who is drifting before the match is lost.
Lecturers and student-support teams can now spot warning signs of academic trouble while there is still time to act, instead of waiting for exam results. That shifts support from reacting to failure toward preventing it, which can improve retention and outcomes. But using student data this way only works if institutions handle privacy and fairness carefully.
The abstract doesn't say how well the framework performs in practice or which machine learning methods work best. It also doesn't define what counts as 'at risk' or how to handle false alarms. It only flags privacy, fairness, transparency, data quality, and inaccurate predictions as issues needing careful consideration.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
AI-Driven Early Identification of At-Risk Students Through Learning Management System Behavioural Analytics — Lawal G. Anand — EdArXiv — doi:10.35542/osf.io/msxnu_v2
Abstract licence: CC BY
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.