Publication

Abstract : Mental fatigue is a widespread issue that significantly affects cognitive performance, emotional well-being, and overall productivity across various professions. This study investigates the implications of occupational mental fatigue using a comprehensive analysis of neurophysiological and biosignal data sourced from the MEFAR (Mental Fatigue Assessment through Neurophysiological and Biosignal Data) dataset. We employed advanced machine-learning techniques, including Random Forest, Logistic Regression, and K-Nearest Neighbors, to classify mental fatigue based on physiological signals such as electroencephalography (EEG), heart rate variability (HRV), and electromyography (EMG). Our findings reveal that the Random Forest model outperformed the other algorithms, achieving an impressive accuracy of 96%, with precision, recall, and F1-score values of 96.55, 95.85, and 96.20%, respectively. In contrast, Logistic Regression demonstrated lower performance metrics, with an accuracy of 63%. The K-Nearest Neighbors model also performed well, achieving 91% accuracy. This research highlights the need for reliable ways to detect mental fatigue at work, showing how machine learning can help improve safety and productivity in challenging jobs.