A Comprehensive Survey Of Artificial Intelligence Techniques For Intelligent Student Learning Analytics And Privacy-Aware Decision Support

Abstract

The exponential development of digital learning platforms has resulted in large volumes of education-related data such as educational records, learning behaviors, cognitive tests, and engagement. The field of Intelligent Student Learning Analytics has become a promising analyze topic as it uses artificial intelligence methods to investigate these heterogeneous data for predicting academic performance, analyzing learner behavior, detecting learner engagement, identifying risks and providing personalized learning. In this survey, review of twenty recently developed methods, which have been reported from 2024 to 2026, related to machine learning, deep learning, explainable artificial intelligence (XAI), federated learning and privacy-preserving learning analytics. A variety of techniques have been utilized in these surveyed methods to enhance educational decision-making and personalized learning; these include but not limited to support vector classification, random forest, artificial neural networks, convolutional neural network, bidirectional long short-term memory, gated recurrent unit, HybridStackNet, SHapley Additive exPlanations (SHAP), Local Interpretable Model-Agnostic Explanations (LIME), differential privacy and federated learning. The survey comprehensively reviews existing techniques, datasets, applications, strengths, weaknesses, etc., emphasizing existing challenges such as lack of generalization of models, heterogeneous nature of educational data, lack of multimodal data integration, privacy issues and lack of interpretable/trustworthy learning analytics. Moreover, the survey discusses existing work of scalable, multimodal and privacy-preserved intelligent learning analytics system that can support reliable educational decision-making.