Slow eye movement detection can prevent sleep-related accidents effectively in a simulated driving task.

A delayed response caused by sleepiness can result in severe car accidents. Previous studies suggest that slow eye movement (SEM) is a sleep-onset index related to delayed response. This study was undertaken to verify that SEM detection is effective for preventing sleep-related accidents. We propose a real-time detection algorithm of SEM based on feature-extracted parameters of electrooculogram (EOG), i.e. amplitude and mean velocity of eye movement. In Experiment 1, 12 participants (33.5 ± 7.3 years) performed an auditory detection task with EOG measurement to determine the threshold parameters of the proposed algorithm. Consequently, the valid threshold parameters were determined, respectively, as >5° and <30° s(-1) . In Experiment 2, 11 participants (32.8 ± 7.2 years) performed a simulated car-following task to verify that the SEM detection is effective for preventing sleep-related accidents. Accidents in the SEM condition were significantly more numerous than in the non-SEM condition (P < 0.01, one-way repeated-measures anova followed by Scheffé's test). Furthermore, no accident occurred in the SEM condition with a warning generated using the proposed algorithm. Results also demonstrate clearly that the SEM detection can prevent sleep-related accidents effectively in this simulated driving task.