0310 Oculometrics Track Performance on the Psychomotor Vigilance Task During Acute Sleep Deprivation

Abstract

Abstract Introduction Sleep deprivation and circadian misalignment impairs human sensorimotor performance and reduces vigilant attention, which increases the potential for errors in occupations that require 24-hour operations. The psychomotor vigilance task (PVT) is the gold-standard measure for evaluating the impact of sleepiness on performance, however, it is not practical to administer in many operational environments, because it only provides a snapshot of performance and requires an individual to focus on the task for several minutes, multiple times over a work shift. As a result, passive, continuous monitoring of sleepiness is desirable for operational environments. The goal of the present study was to determine if complex oculomotor behavioral metrics track PVT performance during sleep deprivation. Methods Twelve healthy adults (mean age 24.8 ± 5.4 years; 6F) maintained a fixed schedule with 8.5 hours in bed for two weeks, during which they abstained from caffeine, alcohol, and other medications, followed by a ~24 hours constant routine laboratory stay. Participants completed the PVT and a radial step-ramp ocular tracking task hourly throughout the study. Twelve oculometrics were derived from smooth pursuit and saccadic eye movements collected through video-oculography and were compared to the PVT and Karolinska Sleepiness Scale (KSS) using linear regression and receiver operating characteristic curves. Results Nine oculometrics spanning pursuit, saccade, and directional motion processing performance correlated with the PVT and KSS (p < 0.05), including: (a) pursuit latency; (b) open-loop pursuit acceleration; (c) proportion smooth; (d) steady-state pursuit gain; (e) saccadic amplitude; (f) saccadic dispersion; (g) saccadic rate; (h) direction asymmetry; and (i) direction noise. Conclusion The oculometrics that we examined exhibited a distinct pattern that tracked PVT performance. Future studies should examine whether these metrics can be extracted through passive monitoring techniques. Support None