Investigation of Sleep Deprivation Effect on Driver #x2019;s Electromyography Signal Features in a Driving Simulator

Background and Objective: Sleep deprivation is an important cause of driver drowsiness. Surface electromyography (sEMG) of the upper arm and the shoulder is an important physiological signal affected by the driver drowsiness. The objective of this paper is to derive the pattern of sleepdeprived drivers #x2019; sEMG. #xD; Materials and Methods: The tests were conducted on 7 men with no sleep disorder aged between 25 and 50 years in a driving simulator. Each subject participated in the tests once without sleep deprivation and another time with two hours of sleep in the 24hour period before the tests. The sEMG signal from the upper arm and shoulder muscles were measured for the mid deltoid, clavicular portion of the pectoralis major, and triceps and biceps long heads. Four features including power spectral kurtosis (SK), mean frequency, absolute amplitude, and root mean square (RMS) were extracted. #xD; Results: The knearest neighbors (kNN) algorithm classifier detected drowsiness with 90% accuracy, 82% precision, 77% sensitivity, and 94% specificity. Driver #x2019;s sleep deprivation can be detected through sEMG signal with 85% accuracy, 80% precision, 70% sensitivity, and 88% specificity. #xD; Conclusion: The sEMG signal amplitude and the frequency content of the sleepdeprived subjects were higher than those of the normal subjects by 37% and 15%, respectively. For the sleepdeprived subjects, muscle contraction did not change much in transition between the last two levels of drowsiness, while the normal subjects experienced 27% drop in this transition. At the last level of drowsiness, the sleepdeprived subjects experienced mental drowsiness without significant change in the muscle contraction level.