A machine learning-based decision support system for temporal human cognitive state estimation during online education using wearable physiological monitoring devices

Over the last decade, there has been a considerable increase in the popularity of online education. As a result, the online learning or e-learning industry has flourished, providing benefits to students, learners, educators, and education experts. Despite the advantages of e-learning, it also has its drawbacks. While e-learning enables students to access the learning materials at their convenience from any location, one of the significant challenges is the lack of monitoring of their level of attention during e-learning sessions. It is challenging to ascertain whether a student is actively engaged in the learning process. To address this issue, we have proposed a decision support system (DSS) based on wearable physiological sensor signals (i.e., Electroencephalogram (EEG) signals) that can inform the instructor whether a student is attentive. For developing DSS, we recorded an EEG-based dataset using a neurosky device, and 100 individuals participated in the study. The learning state is divided into two categories: attentive and inattentive. In this paper, machine learning techniques are employed to integrate the proposed DSS, which can predict, analyze, and validate the student's level of attention or inattention throughout the e-learning session. The findings show that the Support Vector Machine (SVM) approach is the most efficient method for attention prediction, achieving an accuracy of 91.68% compared to logistic regression and ridge regression. Additionally, we examined the frequency bands that were most significant in predicting the learning state, with beta and alpha waves being identified as the key contributors in predicting attention. To further evaluate the data, we use K-means and Hierarchical algorithms to cluster beta and alpha data points. K-means effectively identifies an ideal representative of an attentive or inattentive state. Thus, EEG waves can effectively reveal whether a student is attentive during real-time e-learning sessions, providing a promising approach for providing a valuable tool for decision support in the E-Learning Environment.