Connected Embedded System for Drowsiness Detection Based on a Reconfigurable Set of Features

In this study, we present a new EEG-based drowsiness-detection system using a single EEG channel and IoT technology. The aim of this work is to develop a person-dependent system capable of overcoming interpersonal variability due to aging while sending alert signals to the cloud. We used a set of five features computed from the power spectral density, based on variations in power spectral energy during the transition from wakefulness to drowsiness (stage one of sleep) for each individual. The results demonstrate that the proposed system can accurately detect driver drowsiness, achieving an accuracy of 95% using a reduced set of features and a single differential EEG channel. The main advantage of the proposed system lies in its ability to overcome interpersonal variability while maintaining high detection accuracy. The system was validated using the MIT-BIH Polysomnography dataset, comprising ten subjects.