Abstract: Accurate detection of driver fatigue is essential for improving road safety. This study investigates the effectiveness ofusing multimodal physiological signals for fatigue detection while incorporating uncertainty quantification to enhancethe reliability of predictions. Physiological signals, including Electrocardiogram (ECG), Galvanic Skin Response (GSR),and Electroencephalogram (EEG), were transformed into image representations and analyzed using pretrained deep neural networks. The extracted features were classified through a feedforward neural network, and prediction reliabilitywas assessed using uncertainty quantification techniques such as Monte Carlo Dropout (MCD), model ensembles, andcombined approaches. Evaluation metrics included standard measures (sensitivity, specificity, precision, and accuracy)along with uncertainty-aware metrics such as uncertainty sensitivity and uncertainty precision. Across all evaluations,ECG-based models consistently demonstrated strong performance. The findings indicate that combining multimodal physiological signals, Transfer Learning (TL), and uncertainty quantification can significantly improve both the accuracy andtrustworthiness of fatigue detection systems. This approach supports the development of more reliable driver assistancetechnologies aimed at preventing fatigue-related accidents.

Key words: Fatigue detection, Multimodal physiological signals, Deep transfer learning, Uncertainty-aware learning, Driver monitoring')">Fatigue detection, Multimodal physiological signals, Deep transfer learning, Uncertainty-aware learning, Driver monitoring