A Hybrid Machine Learning Framework for Predicting Academic Performance and Mental Health Risks in Students

Abstract

The increasing prevalence of poor academic performance and mental health challenges among students has created a critical need for intelligent, data-driven intervention systems. This study presented a novel hybrid machine learning framework for the simultaneous prediction of academic performance and mental health risk, addressing the limitations of existing approaches that treat these problems independently. The proposed model employs a stacked ensemble architecture that integrates Support Vector Machine, Random Forest, Artificial Neural Network, and Gradient Boosting as base learners, with Logistic Regression as a meta-learner to enhance predictive accuracy and robustness. An integrated dataset comprising 3,000 student records was constructed from academic data and survey-based psychological assessments, incorporating features such as CGPA, attendance, study habits, stress levels, and sleep quality. To ensure data quality and model reliability, comprehensive pre-processing and validation procedures were applied, including normalization, missing value imputation, and stratified k-fold cross-validation. Experimental results demonstrate that the proposed framework significantly outperforms individual baseline models, achieving an accuracy of 95% for academic performance prediction and 93% for mental health risk classification, with an F1-score of 0.94 and ROC-AUC of 0.96. Furthermore, SHAP-based explainable artificial intelligence techniques were employed to provide interpretable insights into model predictions, revealing key factors influencing student outcomes. The study is limited by its reliance on a single integrated dataset and self-reported psychological measures, which may affect generalizability. This study contributes a scalable and interpretable predictive framework that bridges academic analytics and mental health assessment, offering practical implications for early intervention and decision-making in educational systems.