AI-powered learning analytics dashboards: a systematic review of applications, techniques, and research gaps

This paper presents a Systematic Literature Review (SLR) of Artificial Intelligence (AI)-powered Learning Analytics Dashboards (LADs), focusing on their applications, underlying techniques, and research gaps. A PRISMA-guided search across four major databases identified 21 relevant studies published between 2013 and 2024. The review shows that most AI-powered LADs are used for predicting academic performance, supporting self-regulated learning, and providing teacher-facing insights. Machine learning approaches such as decision trees, ensemble methods (AdaBoost, CatBoost), and Bayesian models are the most frequently employed, with ensemble methods often yielding stronger predictive performance. However, the literature highlights recurring limitations, including small-scale evaluations, limited causal evidence linking predictions to interventions, and weak deployment in real classroom contexts. Ethical considerations, particularly data privacy, bias, and explainability, are also insufficiently addressed. The review concludes by outlining open challenges and proposing directions for future research on scalable, explainable, and ethically grounded AI-powered LADs.