Blink transitions during drowsy periods and threshold inspection of sleep alert

摘要

In this study we propose a drowsiness alert model by using a voltage threshold algorithm in EOG. Here, blink duration and blink interval were calculated as indices of blink activity. Time serial data of blink duration and interval were standardized based on the time section just before the sleep-onset. Blink parameters showed an increased trend toward the sleep-onset point in all eight subjects (reached sleep-onset during watching the movie). The drowsiness alert model demonstrated that all eight subjects were detected as the first-level (light-level) alertness, 6 of 8 subjects were detected as the second-level (severe-level) alertness.