Abstract: Estimating mental fatigue levels during physical exertion is essential for ensuring safety and optimizing performance in sports and other high-demand environments. However, it remains
challenging due to the overlapping physiological signals involved. This study examines the effectiveness of various wearable sensors, including ECG, PPG, and EEG, as well as eye trackers, in estimating mental fatigue levels under conditions of physical fatigue. Multimodal sensor data collected during controlled experiments that combined cognitive tasks with physical exercise were analyzed, with a focus on machine learning models for estimating mental fatigue levels on a 10-class scale. The results show that although EEG is the most accurate single sensor for estimating mental fatigue under physical fatigue, a combination of ECG and eye-tracking achieves higher accuracy (79.16%) with minimal error (MAE=0.52). This combination offers a balance between performance and wearability for practical applications.

Keywords: Fatigue Assessment, Mental Fatigue, Wearable Sensors.