Machine Learning and Student Engagement: The Role of Early Warn-ing Systems in School Dropout Prevention

Abstract

Student engagement, conceptualized as a multidimensional metaconstruct encompassing emotional, cognitive, and behavioral components, is closely linked to dropout risk, academic success, and broader developmental outcomes. Shaped by family, school, peer, and community factors, it serves as a key indicator within prevention efforts. In the Italian context, the growing availability of educational data produced by the digitalization of school systems supports the adoption of Early Warning Systems (EWS) as promising tools for the timely identification of at-risk students. Based on data analytics and machine learning, these tools provide proactive support for monitoring and guiding timely interventions. This contribution provides a research overview on the potential of AI-supported Early Warning Systems in education, examining how these tools can identify patterns of participation, disengagement, and everyday school dynamics. The aim is to explore whether combining algorithmic indicators with contextual qualitative information, professional judgment (human-in-the-loop), and adequate teacher training can enhance the accuracy and usefulness of predictions, while promoting more informed, inclusive, and networked educational practices. Such an approach may enable schools to leverage AI not merely as a monitoring tool, but as a catalyst for targeted actions that support student participation, agency, and long-term well-being.