Exploring the Relationship between Neural Mechanism and Detection in Mental Fatigue by Genetic Algorithm and Hierarchical Clustering

Abstract

The mental fatigue affects the state of one's daily life easily, therefore, understanding the neural mechanisms of mental fatigue and better detection of it have been the focus of many researchers. Quit a few previous studies have found EEG indicators and high-precision detection methods related to mental fatigue, however, how to combine these EEG indicators with detection methods for better detection remains to be solved. To classify mental fatigue based on EEG features, our previous research, which adopted GA-SVM method, have demonstrated the optimal channels are mainly distributed in the prefrontal, occipital and temporal lobes, and the optimal channel number is 5. Here, we further explored the question by developing a new method combining genetic algorithm and hierarchical clustering to study the mental fatigue caused by visual search. Our results suggest that the optimal EEG features for assessing fatigue state vary from person to person, while the corresponding optimal channel positions are consistent. The channels with the largest changes in EEG features are mainly distributed in the frontal lobe, followed by the temporal lobe and a small area of the occipital lobe, while the corresponding regions of the almost all parietal lobe and part occipital lobe show little changes in EEG features during fatigue. Current study shows that the optimal EEG features of different individuals are different in the mental fatigue detection, and they need to be determined separately, but only a few of the same channels can be used to achieve the better detection.