부산대학교 도서관

Electrodermal Activity (EDA) is a broadly-investigated physiological signal, whose behaviour is connected to nervous system arousal. Such system, indeed, influences the properties of the skin, producing a measurable electrical signal. Among the possible applications of such measurements, several studies have correlated the signal behaviour to engagement during mental and physical tasks, and the subjects' response to specific multimodal stimuli. Also due to the possibility of performing remote assessment and rehabilitation, telemedicine applications are gaining ground in the healthcare system. However, acceptance and engagement, hence continuity of usage, still remain significant obstacles. Therefore, it would be highly beneficial to verify, through objective measures, if these solutions are actually providing a sufficient stimulation to properly engage subjects while playing. This study investigates the possibility of employing EDA in the automatic recognition of different levels of user engagement, while playing a motor-cognitive exergame specifically designed for this purpose. Preliminary results, obtained on a cohort of 25 healthy subjects, seem to confirm that features extracted from EDA analysis are significant and able to train supervised classifiers, achieving high accuracy and precision in the engagement recognition problem.