Sensitivity and specificity of the driver sleepiness detection methods using physiological signals: a systematic review
Article
Article Title | Sensitivity and specificity of the driver sleepiness detection methods using physiological signals: a systematic review |
---|---|
ERA Journal ID | 13370 |
Article Category | Article |
Authors | Watling, Christopher N. (Author), Hasan, Md Mahmudul (Author) and Larue, Gregoire S. (Author) |
Journal Title | Accident Analysis and Prevention |
Journal Citation | 150, pp. 1-11 |
Article Number | 105900 |
Number of Pages | 11 |
Year | 2021 |
Publisher | Emerald |
Place of Publication | United Kingdom |
ISSN | 0001-4575 |
1879-2057 | |
Digital Object Identifier (DOI) | https://doi.org/10.1016/j.aap.2020.105900 |
Web Address (URL) | https://www.sciencedirect.com/science/article/pii/S0001457520317206 |
Abstract | Driver sleepiness is a major contributor to road crashes. A system that monitors and warns the driver at a certain, critical level of arousal, could aid in reducing sleep-related crashes. To determine how driver sleepiness detection systems perform, a systematic review of the sensitivity and specificity outcomes was performed. In total, 21 studies were located that met inclusion criteria for the review. The range of sensitivity outcomes was between 39.0-98.8% and between 73.0-98.9% for specificity outcomes. There was considerable variation in the outcomes of the studies employing only one physiological measure (mono-signal approach), whereas, a poly-signal approach with multiple physiological signals resulted in more consistency with higher outcomes on both sensitivity and specificity metrics. Only six of the 21 studies had both sensitivity and specificity outcomes above 90.0%, which included mono- and poly-signal approaches. Moreover, increases in the number of features used in the sleepiness detection system did not result in higher sensitivity and specificity outcomes. Overall, there was considerable variability between the studies reviewed, including measures of ground truth, the features employed and the machine learning approach of the systems. A critical need for progressing any system is a revalidation of the system on a new sample of users. These aspects indicate considerable progress is needed with physiological-based driver sleepiness systems before they are at a sufficient standard to be deployed on-road. |
Keywords | fatigue, drowsiness, driving, features, machine learning, ground truth, physiological sleepiness |
ANZSRC Field of Research 2020 | 420604. Injury prevention |
520206. Psychophysiology | |
520404. Memory and attention | |
Public Notes | File reproduced in accordance with the copyright policy of the publisher/author. |
Byline Affiliations | Queensland University of Technology |
Institution of Origin | University of Southern Queensland |
https://research.usq.edu.au/item/q6zy2/sensitivity-and-specificity-of-the-driver-sleepiness-detection-methods-using-physiological-signals-a-systematic-review
Download files
Submitted Version
ePRINTS Watling etal 2021 Sensitivity and specificity of the driver.pdf | ||
File access level: Anyone |
84
total views299
total downloads1
views this month1
downloads this month