Enhancing Driver Stress Detection through Multimodal Integration of Eye Tracking and Physiological Signals

Widayat, Tri Agung
Mintje, Quirina Ariantji Patrisia
Yosepha, Sri Yanthy

Logistica : Journal of Logistic and Transportation
Vol. 3 No. 3 (2025): July 2025

Driver stress poses significant risks to traffic safety, impairing attention, decision-making, and reaction time. Traditional monitoring methods often lack sensitivity. This study proposes and validates a novel multimodal framework that integrates synchronized eye-tracking and physiological data to significantly enhance the sensitivity and real-time accuracy of driver stress detection, addressing limitations of earlier unimodal approaches. Thirty licensed drivers participated in simulated driving tasks under baseline and stress-induced conditions. Eye-tracking metrics (pupil diameter, fixation duration, blink rate) and physiological signals (heart rate, skin conductance, heart rate variability) were collected. Data were synchronized and analyzed using Linear Discriminant Analysis (LDA) and other machine learning models to classify stress conditions. Under stress, pupil dilation increased by 20%, blink rate rose by 35%, and gaze spread narrowed, indicating visual tunneling. Physiologically, heart rate increased by 17%, skin conductance by 31%, and HRV decreased by 19%. The combined multimodal model achieved 91.4% classification accuracy, outperforming unimodal approaches. These results align with previous research showing that multimodal systems provide more reliable stress detection by integrating visual and autonomic markers. The findings highlight the system’s potential for real-time applications in Driver Monitoring Systems (DMS). Multimodal integration of eye-tracking and physiological signals enhances the sensitivity and reliability of driver stress detection. This approach offers a foundation for intelligent, adaptive DMS capable of improving road safety. Future work should focus on real-world validation and ethical implementation strategies. These findings demonstrate that multimodal integration provides a more comprehensive understanding of driver stress through complementary visual and autonomic indicators. The proposed framework forms a foundation for intelligent, adaptive Driver Monitoring Systems (DMS) capable of real-time stress recognition and proactive safety intervention.