Affiliations: [a]
Department of Occupational Health Engineering, School of Public Health, Tehran University of Medical Sciences, Tehran, Iran
| [b]
School of Data Science, Nagoya City University, Nagoya, Japan
| [c] Virtual Reality Laboratory, K.N. Toosi University of Technology, Tehran, Iran
| [d]
Department of Computer Engineering and Information Technology, Amirkabir University of Technology, Tehran, Iran
| [e]
University of Pittsburgh, Pittsburgh, PA, USA
Correspondence:
[*]
Address for correspondence: Adel Mazloumi, Department of Occupational Health Engineering, School of Public Health,
Tehran University of Medical Sciences, Tehran, Iran. E-mail: amazlomi@tums.ac.ir.
Abstract: BACKGROUND:The transition from alertness to drowsiness can cause considerable changes in the respiratory system, providing an opportunity to detect driver drowsiness. OBJECTIVE:The aim of this study was to determine which respiratory features indicate driver drowsiness and then use these features to classify the level of drowsiness and alertness. METHODS:Twenty male students (mean age 25.6±2.41 years) participated in the study using a driving simulator, and eight features, including expiration duration (ED), inspiration duration (ID), peak-to-peak amplitude (PA), inspiration-to-expiration time ratio (I/E ratio), driving, timing, respiration rate (RR), and yawning, were extracted from the respiratory signal generated by abdominal motions using a belt equipped with a force sensor. RESULTS:All eight features were statistically significant at the significance level of 0.05. Drowsiness can be detected using respiratory features with 88% accuracy, 82% precision, 86% recall, and an 90% F1 score. CONCLUSION:The findings of this study may be useful in the development of driver drowsiness monitoring systems based on less intrusive respiratory signal analysis, particularly for specific process automation applications when vehicle control is not in the hands of the driver.
