Tracking intermediate performance of vigilant attention using multiple eye metrics

Abstract

Vigilance deficits account for a substantial number of accidents and errors. Current techniques to detect vigilance impairment measure only the most severe level evident in eyelid closure and falling asleep, which is often too late to avoid an accident or error. The present study sought to identify ocular biometrics of intermediate impairment of vigilance and develop a new technique that could detect a range of deficits in vigilant attention (VA). Sixteen healthy adults performed well-validated Psychomotor Vigilance Test (PVT) for tracking vigilance attention while undergoing simultaneous recording of eye metrics every 2 hours during 38 hours of continuous wakefulness. A novel marker was found that measured VA when the eyes were open—the prevalence of microsaccades. Notably, the prevalence of microsaccades decreased in response to sleep deprivation and time-on-task. In addition, a novel algorithm for detecting multilevel VA was developed, which estimated performance on the PVT by integrating the novel marker with other eye-related indices. The novel algorithm also tracked changes in intermediate level of VA (specific reaction times in the PVT, i.e. 300–500 ms) during prolonged time-on-task and sleep deprivation, which had not been tracked previously by conventional techniques. The implication of the findings is that this novel algorithm, named “eye-metrical estimation version of the PVT: PVT-E,” can be used to reduce human-error-related accidents caused by vigilance impairment even when its level is intermediate.

eye movements, neurobehavioral performance, sleep deprivation, biomarkers, computerized analysis, fatigue, PERCLOS, blink, eyelid movements, pupil, Bayesian inference