Data-Driven Learning Analytics and Artificial Intelligence in Higher Education: A Systematic Review

Abstract

The responsible integration of Artificial Intelligence in Education (AIED) offers a strategic opportunity to align learning environments with the principles of Society 5.0, fostering human–technology synergy in support of quality education and social well-being. This study presents a systematic review of 36 peer-reviewed articles (2021–2025) focused on educational appli-cations that employ learning analytics (LA) through data-driven approaches and integrate machine learning (ML) models as part of their empirical evidence. Each study was analyzed according to three key dimensions: the context of AIED application, the data-driven approach adopted, and the ML model implemented. The findings reveal a persistent disconnect between the AI models employed and the available educational data, which in many cases are limited to access logs or manually recorded grades that fail to capture deeper cognitive processes. This limitation constrains both the effective training of ML models and their pedagogical utility for delivering meaningful interventions such as personalized learning pathways, real-time feedback, early detection of learning difficulties, and monitoring and visualization tools. Another significant finding is the absence of psychopeda-gogical frameworks integrated with quality standards and data governance, which are essential for advancing prescriptive and ethical approaches aligned with learning goals. It is therefore recommended that educational leaders foster AIED applications grounded in data governance and ethics frameworks, ensuring valid and reliable metrics that can drive a more equitable and inclusive education.