Identifying periods of drowsy driving using EEG.

Drowsy driving is a significant contributor to death and injury crashes on our nation's highways. Predictive neurophysiologic/physiologic solutions to reduce these incidences have been proposed and developed. EEG based metrics were found to be promising in initial studies, but remain controversial in their efficacy, primarily due to failures to develop replication studies within the simulation settings used for development, and real-world validation. This analysis sought to address these short comings by assessing the utility of the B-Alert algorithms, in a replication study of driving and drowsiness. Data were collected on the National Advanced Driving Simulator from 72 volunteer drivers exposed to three types of roadways at three times of day representing different levels of drowsiness. EEG metrics, collected using the B-Alert X10 Wireless Headset were evaluated to determine their utility in future predictive studies. The replication of the B-Alert algorithms was a secondary focus for this analysis, resulting in highly variable start times within each time of day segment, leading to EEG data being confounded by the diurnal variations that occur in the basal EEG signal. Regardless of this limitation, the analysis revealed promising outcomes. The EEG based algorithms for sleep onset, drowsiness, as well as fatigue related power spectral bandwidths (i.e. lateral central, and parietal alpha) varied with time of day of the drives. Interestingly, EEG metrics of cognitive workload were also sensative to the terrain of the drives. The replicaiton of the B-Alert algorithms were a secondary focuse in the study design, Taken together, these data indicate great potential of carefully designed studies to utilize neurophysiologic metrics to identify time of day and task and road conditions that may be at greatest risk during fatigued/drowsy periods.