Multinomial Logistic Regression Model for Predicting Driver's Drowsiness Using Behavioral Measures

AbstractThe aim of this study was to explore the effectiveness of behavioral evaluation measures for predicting drivers’ subjective drowsiness. Behavioral measures included neck vending angle (horizontal and vertical), back pressure, foot pressure, COP (Center of Pressure) movement on sitting surface, and tracking error in driving simulator task. Drowsy states were predicted by means of the multinomial logistic regression model where physiological and behavioral measures and subjective evaluation of drowsiness corresponded to independent variables and a dependent variable, respectively. First, we compared the effectiveness of two methods (correlation coefficient-based method and odds ratio-based method) for determining the order of entering behavioral measures into the prediction model. It was found that the prediction accuracy did not differ between both methods. Second, the prediction accuracy was compared among the numbers of behavioral measures. The prediction accuracy did not differ among four, five, and six behavioral measures, and it was concluded that entering at least four behavioral measures into the prediction model is enough to achieve higher prediction accuracy. Third, the prediction accuracy was compared between the strongly drowsy and the weakly drowsy group. The prediction accuracy differed between the two groups, and the proposed method was effective (the prediction accuracy was significantly higher) especially under the condition where drowsiness was induced to a larger extent