AI-Powered Student Dropout Prediction and Intervention System Using Gemini AI and Supabase Integration

Abstract

This project presents a web-based intelligent system designed to predict potential student dropouts using Artificial Intelligence (AI) and provide automated, personalized counselling to at-risk students. The proposed model integrates predictive analytics, real-time data processing, and generative AI-based de- cision support. The system architecture combines a React-based frontend, Supabase backend, and Gemini AI agent for natural language interaction and adaptive feedback. It analyzes both academic (attendance, grades, performance) and non-academic parameters (engagement, sentiment, socio-economic context) to calculate dropout risk scores and suggest tailored interventions. By combining AI prediction with human feedback loops, the system empowers educators to identify, understand, and assist students before disengagement occurs, significantly improving student retention and academic success rates.

Introduction

Student dropout remains a major global challenge, with over one-third of undergraduates failing to complete their programs due to academic, motivational, or socio-economic factors. Traditional academic monitoring systems are largely reactive, identifying problems only after significant performance decline. Recent advances in AI and machine learning enable predictive dropout detection, but most existing systems focus only on prediction accuracy and lack personalized, real-time intervention and counselling.

This research proposes an AI-powered, integrated dropout prevention system that combines predictive analytics with conversational AI. The system uses machine learning models to analyze multidimensional data—academic performance, behavioral patterns, and socio-economic factors—to generate a Dropout Probability Score (DPS). Students are categorized into low-, moderate-, or high-risk groups, triggering automated interventions.

A key innovation is the integration of Gemini AI, which provides personalized counselling, motivational feedback, study recommendations, and alerts for teacher follow-up. Real-time data handling is enabled through a Supabase backend and a React-based dashboard, allowing teachers to monitor progress, record feedback, assign customized tasks, and collaborate on student support.

The proposed framework addresses major research gaps by combining real-time analytics, structured and unstructured data (including sentiment analysis of feedback), and interactive AI-driven counselling. Overall, the system aims to shift dropout management from a reactive to a proactive approach, improving student retention through timely, data-driven, and personalized academic support.

Copyright

Copyright © 2025 Mohammed Zaid Ejaj Pathan , Bharadwaj Premising Patil, Om Ganesh Devkar, Prof. Nilesh S. Ahire. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.