Traffic injury prevention · 2026
Modeling behavioral indicators for driver drowsiness detection: a simulator-based study.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Azaddel Rahele, Rezaei Mahdi, Sadeghi-Bazargani Homayoun, Rasoulzadeh Yahya
Аннотация
OBJECTIVE: Driver drowsiness is a critical factor in road accidents. This study aimed to model behavioral indicators of drowsiness using a driving simulator to support noninvasive detection systems.
METHODS: Twenty-four participants completed simulated driving tasks under varying alertness levels. Behavioral metrics including eye-blinking frequency, head movement acceleration, and eye movement variability were recorded. Drowsiness classification was performed using supervised Partial Least Squares Discriminant Analysis (PLS-DA), with the Karolinska Sleepiness Scale as the reference standard. Preprocessing steps included general mean-centering, pairwise mean-centering, and unit variance scaling.
RESULTS: Blink frequency significantly increased with drowsiness, while fluctuations in head movement acceleration and eye movements also rose, indicating reduced alertness. Derived variables such as Corrected Turning Ratio (CTR) showed predictive relevance, whereas angular velocity of head movement was not statistically significant. Model evaluation demonstrated strong performance (ROC AUC = 0.935), with low misclassification rates and acceptable residual normality.
CONCLUSIONS: Behavioral metrics provide practical predictive value for noninvasive drowsiness detection. Although limited by sample size, the model achieved meaningful separation between alertness and drowsiness. Future studies should expand sample size, incorporate additional behavioral indicators, and validate findings externally to strengthen predictive performance.
METHODS: Twenty-four participants completed simulated driving tasks under varying alertness levels. Behavioral metrics including eye-blinking frequency, head movement acceleration, and eye movement variability were recorded. Drowsiness classification was performed using supervised Partial Least Squares Discriminant Analysis (PLS-DA), with the Karolinska Sleepiness Scale as the reference standard. Preprocessing steps included general mean-centering, pairwise mean-centering, and unit variance scaling.
RESULTS: Blink frequency significantly increased with drowsiness, while fluctuations in head movement acceleration and eye movements also rose, indicating reduced alertness. Derived variables such as Corrected Turning Ratio (CTR) showed predictive relevance, whereas angular velocity of head movement was not statistically significant. Model evaluation demonstrated strong performance (ROC AUC = 0.935), with low misclassification rates and acceptable residual normality.
CONCLUSIONS: Behavioral metrics provide practical predictive value for noninvasive drowsiness detection. Although limited by sample size, the model achieved meaningful separation between alertness and drowsiness. Future studies should expand sample size, incorporate additional behavioral indicators, and validate findings externally to strengthen predictive performance.