This study aimed to assess steering entropy as a measure of decrements indriving performance caused by microsleeps. Microsleeps are brief, unintended episodes of loss of attention that last 3-14 seconds. These episodes, which are frequent in drivers with sleep disorders, can be long enough to impact steering performance and are particularly disruptive when driver action is imperative, as when driving around curved highway segments. Steering entropy is a driver-centered performance measure that can detect drivers' corrective responses to situations when the vehicle state falls outsidethe driver's expectations. This study tests the hypothesis that steering entropy is an indicator of increased erratic steering behavior during microsleep episodes in drivers with obstructive sleep apnea/hypopena syndrome (OSAHS). Twenty-four drivers with OSAHS were used in this study and their electroencephalography (EEG) defined microsleep (cases) and non-microsleepepisodes (crossover control) were compared using a case-crossover method.The performance measure, steering entropy, was calculated from a time-series history of steering angle data. Steering entropy was compared for eachmicrosleep in the three-second interval both immediately preceding and immediately following each microsleep. Results showed that steering entropy was higher on curves during microsleeps and post microsleeps when comparedto straight road segments and the no-workload baseline condition. This suggests that steering entropy can capture erratic steering behavior, allowing us to better understand how drivers correct for previous steering errors.
Abstract