Causal brain network analysis of driving fatigue based on generalized orthogonalized partially directed coherence

Abstract

Driving fatigue is a serious threat to driving safety. Therefore, it is of great significance to accurately detect driving fatigue. In this study, the generalized orthogonal partial directed coherence (gOPDC) algorithm, which measures the time–frequency domain interaction of electroencephalogram (EEG) signals, was used to accurately estimate the connectivity between cortical channels. The causal brain network of driver continuous driving is constructed. The results show that the clustering coefficient and global efficiency tend to decrease with the increase in driving time. Causal information flow in the left prefrontal, parietal, occipital regions and the right posterior frontal region increased significantly when subjects transitioned from awake to fatigued, while causal information flow in the right prefrontal, parietal, occipital regions and the left posterior frontal region decreased mutually significantly. Compared with the traditional driving fatigue algorithm, the accuracy of the method used in this paper is higher than the traditional methods.