An analysis on driver drowsiness based on reaction time and EEG band power.

Falling asleep during driving is a serious problem that has resulted in fatal accidents worldwide. Thus, there is a need to detect driver drowsiness to counter it. This study analyzes the changes in the electroencephalography (EEG) collected from 4 subjects driving under monotonous road conditions using a driving simulator. The drowsiness level of the subjects is inferred from the time taken to react to events. The results from the analysis of the reaction time shows that drowsiness occurs in cycles, which correspond to short sleep cycles known as `microsleeps'. The results from a time-frequency analysis of the four frequency bands' power reveals differences between trials with fast and slow reaction times; greater beta band power is present in all subjects, greater alpha power in 2 subjects, greater theta power in 2 subjects, and greater delta power in 3 subjects, for fast reaction trials. Overall, this study shows that reaction time can be used to infer the drowsiness, and subject-specific changes in the EEG band power may be used to infer drowsiness. Thus the study shows a promising prospect of developing Brain-Computer Interface to detect driver drowsiness.