Artificial Intelligence-Driven Personalized Learning Models for Special Education

Abstract

Learning personalized through the use of AI is gaining importance in special education since there is a need for learners who require special attention to receive personalized learning. Predictive machine learning can be used to analyze learning patterns from participation data. This study aimed to develop and evaluate AI-driven personalized learning models for predicting academic outcome categories using learner-related educational variables. A quantitative machine learning design was applied to a dataset containing 14,003 observations and 16 variables. After preprocessing, the ExamScore variable was removed to prevent target leakage. The data were divided into training and testing subsets using an 80:20 stratified split. Decision Tree and Random Forest classifiers were trained and evaluated using accuracy, precision, recall, weighted F1-score, confusion matrices, ROC-AUC analysis, and feature importance. The Random Forest classifier substantially outperformed the Decision Tree model. Random Forest achieved 84.33% accuracy, 85.04% precision, 84.33% recall, and an 84.29% weighted F1-score, while Decision Tree achieved 32.42% accuracy and a 29.49% weighted F1-score. The Random Forest model also produced strong multiclass ROC-AUC performance, with a macro-average AUC of 0.961. Assignment completion, attendance, study hours, online course participation, and age were the most influential predictors. The ensemble-based method will help to apply personalized AI learning based on detecting engagement patterns of students and predicting their performance. It can be concluded that it is possible to apply this learning technique in special education programs.