Just valid or even accurate? Determine the measurement accuracy of the pupillographic sleepiness test by applying self and observer ratings

Objective. The purpose of the present study was to provide validation and accuracy data for the pupillographic sleepiness test (PST), on the one hand, by applying self and observer ratings and, on the other hand, by fusioned self and observer ratings as a sleepiness reference value.Methods. A total of 30 healthy women and men participated in a partial sleep deprivation study (20.00–04.00 h) and PST measurements were conducted every 2 h for a total of 113 PST measurements. Karolinska Sleepiness Scale (KSS)-based self-reports and five video-based observer ratings of drowsiness (ORD) were measured immediately before the PST in order to provide reliable reference sleepiness values.Results. PST parameters (pupil unrest index, power of frequency) correlated significantly with the sleepiness validation criteria the KSS and ORD used in this study. Fusioned reference values obtained from one self-report and observer ratings showed a correlation of r = 0.54, a mean absolute percentage error of 1.58 KSS points, and an error of 35%.Conclusion. Our results indicate the moderate validity of the PST. Furthermore, the proposed sleepiness reference value might serve as a feasible intermediate solution to estimate sleepiness in the sense of a reference (“quasi-ground truth”) value. This might be true especially for within-subject designs with a focus on the time course of sleepiness. Moreover, the results might show the necessity to recalculate the thresholds of the current PST categories of sleepiness severity.

Access to full text restricted pending copyright check.