Multi‐modal‐Sensing System for Detection and Tracking of Mind Wandering

Summary

This chapter delves deeper into the phenomenon of mind wandering by employing dynamic stimuli, specifically video content, as opposed to static stimuli. It presents a novel approach where a proof‐of‐concept wearable multisensory system was developed and integrated with machine learning techniques to improve the detection of students' learning and concentration levels. A decision‐level fusion method was devised to aggregate the confidence levels from all sensing techniques, enhancing the overall detection accuracy. To evaluate the effectiveness of the wearable multisensory device, a pilot study was conducted with 10 participants, comprising university students of both genders aged between 21 and 30. Two distinct machine learning methods, support vector machine (SVM) and gated recurrent unit (GRU), were employed for the classification models. The results demonstrated that the fusion of sensor data using SVM and GRU achieved notable accuracies of 86.53% and 89.86%, respectively.