Development of an Evaluation Method for a Driver’s Cognitive Workload Using ECG Signal

Abstract

A driver’s high cognitive workload can be a reason to increase car accident incidence because high cognitive workload makes the driver confused while driving. So evaluation methods for drivers’ cognitive workload using electrocardiography (ECG) has been researched as a method to support the safe driving through alarms in advance. However, there is a limitation that the measures to quantify ECG data have been applied to all the drivers equally

consequently, the accuracy of cognitive workload evaluation was low. But if heartbeat characteristics of all the drivers are analyzed and the optimal analysis methods are determined individually, the accuracy of cognitive workload evaluation would be higher.The present study is intended to develop a judgment method of optimal cognitive workload analysis conditions considering the individual heartbeat characteristics and evaluate the effectiveness of the method developed. First, ECG quantification measures and factors for real-time analysis were defined through literature review to judge individual optimal cognitive workload analysis conditions based on ECG. Mean interbeat interval (IBI), standard deviation of N-N intervals (SDNN), root mean of sum of squared differences (RMSSD), and root mean square error (RMSE) are determined as the ECG quantification measures in the study. And window span and update rate were defined as the factors for real-time analysis and 20, 30, and 40 seconds of window span and 1, 2, and 3 seconds of update rate were defined as the factor levels.Second, the judgment method of individual optimal cognitive workload analysis conditions was developed. The defined factor levels were combined to 36 cognitive workload analysis conditions (4 ECG quantification measures × 3 window spans × 3 update rates). The individual optimal cognitive workload analysis conditions among the 36 conditions were determined through area under the ROC curve (AUC) analysis using collected ECG data from n-back task (secondary task).Lastly, the effectiveness test for the judgment method developed of the individual optimal cognitive workload analysis conditions was conducted using collected ECG data from the simulator experiment. In the results of 7 (47%) of 15 participants, the judgment method of the individual optimal cognitive workload analysis conditions was effective. On the other hand, there was the limitation that the results of 8 participants (53%) were not effective

therefore, the method will be completed by adding cognitive workload analysis conditions to cover more various heartbeat characteristics in further studies.In summary, the developed judgment method of the individual optimal cognitive workload analysis conditions provides about 47% of effectiveness. And the method will be completed to apply it to all the drivers by understanding their individual heartbeat characteristics.