Prediction of automobile driver sleepiness (1st report, rating of sleepiness based on facial expression and examination of effective predictor indexes of sleepiness)

Abstract

An automobile simulator experiment was carried out in order to develop a method of prediction of sleepiness based on mainly physiological parameters. Twelve subjects participated in the study. First, criteria variables of sleepiness were investigated since such variables have not been clear in many previous studies. Rated sleepiness based on facial expression was examined as a criteria variable in this research. Rated sleepiness by two examiners gave high correspondence (0.755), and mean correlation coefficient between this rated sleepiness and the subjects' self rating sleepiness was also high (0.794). These results show that this rated sleepiness is considered to be reliable and valid as a criteria variable. Second, both physiological parameters and the driving performances were adopted as explanatory variables. Cross correlation analysis between the rated sleepiness and these parameters showed that α/β at posterior-midline (Pz), blink duration, the rate of long blink and heart rate were relevant as general indexes of sleepiness. Methods of prediction of sleepiness based on these results are mentioned in our second report.