Abstract
BACKGROUND: Driver fatigue detection is crucial for traffic safety. Electroencephalography (EEG) signals, which directly reflect the human mental state, provide a reliable approach for identifying fatigue. OBJECTIVE: This study aimed to investigate the effectiveness of EEG microstate analysis in detecting driver fatigue by analyzing variations in microstate features between normal and fatigued states. MATERIAL AND METHODS: This analytical study aimed to develop a supervised machine learning approach for driver fatigue detection using EEG microstate features. EEG data were collected from 10 individuals in both normal and fatigued states. Microstate analysis was performed to extract key features, including duration, occurrence, coverage, and Microstate Mean Power (MMP), from four types of microstates labeled A, B, C, and D. These features were then used as inputs to train and test a Support Vector Machine (SVM) for classifying each EEG segment into either normal state or fatigue state. RESULTS: The classification achieved high accuracy, particularly when combining MMP and occurrence features. The highest accuracy recorded was 98.77%. CONCLUSION: EEG microstate analysis, in combination with SVM, proves to be an effective method for detecting driver fatigue. This approach can be utilized for real-time driver monitoring and fatigue alert systems, enhancing road safety.