The work explores the integration of behavioral analytics and artificial intelligence (AI) techniques, specifically Graph Convolutional Networks (GCNs) and deep learning, to predict student academic performance. By analyzing study habits, assignment submissions, gaming behavior, and exam scores, the framework bridges traditional evaluation methods and behavioral insights. Results demonstrate that a GCN model outperforms other machine learning approaches, achieving a Mean Squared Error (MSE) of 117.97. The findings highlight the potential of machine learning to enhance educational strategies, identify at-risk students, and provide actionable insights for data-driven educational improvements. © 2025 Elsevier B.V., All rights reserved.