Ai-based predictive modeling of student performance in Moodle: A case study from the COLOURS alliance

European university alliances aim to enhance student performance and enable personalized learning through the integration of digital platforms and intelligent tools that support teaching and learning processes. This paper explores the application of Artificial Intelligence (AI) to predict students at risk of academic failure by analyzing their activity within Moodle, a widely used e-learning platform. Our case study is the European university alliance COLOURS and its Moodle platforms across each partner university, but also the interoperable infrastructures set on Alliance level (be it Moodle of Moodles or specially designated aggregators and portals. The analysis considers multiple indicators, including the number of completed assignments, hours spent studying, participation in discussion forums, attendance in learning activities, and engagement with digital resources such as learning materials, quizzes, and simulations. We intend to incorporate quantitative data first, but at later stages of the research, also qualitative aspects, as well as diverse contexts. For the purposes of the COLOURS alliance analysis, a Random Forest machine learning model is implemented in Python using Google Colab to analyze Moodle activity data and predict at-risk students. Student activity data are collected from Moodle through a Learning Record Store (LRS), which ensures standardized xAPI statements and reliable data extraction. This approach leverages the rich dataset collected in Moodle and the predictive capabilities of AI to support early risk detection and personalized learning recommendations. The expected outcome is the early identification of at-risk students, enabling timely interventions and contributing to the development of more effective, personalized learning strategies that enhance academic achievement.