abstract
- The increasing prevalence of hybrid education has underscored the need for strategies to evaluate student concentration across both in-person and remote settings. Within Human-Computer Interaction (HCI) contexts, monitoring attention during various learning tasks has been identified as a critical factor influencing academic outcomes. Although prior research has demonstrated the utility of electroencephalography (EEG) for assessing mental states, the application of low-cost, portable EEG devices for concentration monitoring in educational environments remains limited.This study aimed to examine the feasibility of using EEG signals recorded from a Muse 2 Headband to distinguish between tasks requiring different levels of concentration. A total of forty-one participants completed a series of digital tasks while EEG data were collected. The dataset was processed and analyzed using three machine learning algorithms: Random Forest, Support Vector Machine, and KNearest Neighbors. The objective was to classify concentration versus non-concentration states based on EEG features and selfreported focus levels. The Random Forest model achieved the highest classification accuracy, outperforming comparable methods and supporting the validity of using the Muse 2 Headband as a real-time, cost-effective tool for concentration assessment in HCI-driven educational scenarios. These findings suggest that portable EEG devices may serve as effective instruments for tracking concentration in hybrid learning environments. Their integration could offer timely feedback to students and educators, contributing to improved academic performance. Future work will explore multimodal approaches by combining EEG data with other physiological signals to enhance the robustness of concentration assessment models.