Researcher Essa E. Almazroei from the Learning Design and Technology Department at the College of Education, University of Jeddah, has published a pioneering scientific study aimed at enhancing student academic performance in online learning environments. The study introduces an innovative framework that integrates robust machine-learning algorithms with Explainable Artificial Intelligence (XAI) techniques. Study Details and Methodology To develop a highly accurate predictive model, the researcher integrated seven diverse educational datasets into a unified analytical pipeline. These data sources included: - Learner demographics. - Assessment history and cumulative grades. - Virtual Learning Environment (VLE) engagement levels and interaction traces. - Course registration and withdrawal patterns. Four supervised machine learning algorithms were trained and evaluated during the research: Logistic Regression, Random Forest, XGBoost, and Multi-Layer Perceptron (MLP). Key Findings and Academic Impact The study demonstrated high efficiency in identifying students at risk of academic underperformance at an early stage. The primary outcomes of the research include: - The XGBoost model emerged as the strongest performer, achieving an outstanding accuracy rate of 95.04%. - Advanced XAI techniques, specifically SHAP and LIME, were employed to ensure the transparency and interpretability of the system's predictive decisions. - Cumulative assessment performance, withdrawal patterns, and engagement intensity were identified as the most influential predictors of student success. - The research provides a reliable "early-warning" system that enables educational institutions to deliver proactive, targeted academic interventions and customized student support. This scientific paper represents a significant step forward in leveraging large-scale educational data to improve institutional decision-making, enhance personalized academic advising, and reduce dropout rates in digital learning systems.