This paper explores how to utilize multi-biometric techniques to detect distraction due to smartphone use during computer-based online learning, a task that requires sustained attention. While it can be applied to a variety of fields such as autonomous driving, our study focuses on factors that impede learner engagement, such as intrinsic factors (e.g., motivation), system-related factors (e.g., course design), and situational factors (e.g., smartphone use). While existing learning platforms lack detailed behavioral data, multi-modal learning analysis (MMLA) and biometric sensors can provide new insights into learner attention. In this study, we propose an AI-based approach that utilizes physiological signals and head pose data to detect smartphone use. Our results show that while single biometric signals such as EEG or heart rate have limited accuracy, head pose alone achieves 87% accuracy. A multi-modal model that combines all signals achieves 91% accuracy, highlighting the benefits of integration. Finally, we discuss the implications of supporting such models in real time in online learning environments and Limitations.
