Estimation of driver sleepiness based on EEG analysis

IEEE Transactions on Aerospace and Electronic Systems, 2019

Drowsiness detection has a significant importance in aviation industry. Electroencephalogram (EEG) has been extensively studied to characterize drowsiness, nevertheless, its behavior with respect to accurately annotated drowsy periods remains to be investigated. At present, Karolinska Sleepiness Scale (KSS) and Psychomotor Vigilance Task (PVT) are widely used references for subjective drowsiness analysis. However, their practical application in real time monitoring of alertness is limited. To address this limitation, we combined Seeing Machines Driver Monitoring System (DMS) and electrooculogram (EOG) for localization of microsleep (MS) events and studied EEG spectral behavior during MS events. EEG, EOG, and facial behavior data were recorded simultaneously from sixteen commercially-rated pilots during simulated flight. Relative spectral power in delta (0.5-4 Hz), theta (4-8 Hz), alpha (8-13 Hz), and beta (13-30 Hz) in frontal, central, temporal, parietal, and occipital brain regions were analyzed. Compared to baseline, delta power reduced during MS (p<0.05 in all regions), alpha power increased during MS (p<0.001 in all regions), while theta and beta powers did not change (p>0.05). The research findings highlight the capability of EEG delta and alpha spectrum towards characterizing MS events, therefore, with consideration to user acceptability, application towards drowsiness detection is plausible via EEG electrodes embedded in the typical aviation headset.

Periodogram and other spectral power estimation methods are established in quantitative EEG analysis. Their outcome in case of drowsy subjects fulfilling a sustained attention task is difficult to interpret. Two novel kind of EEG analysis based on pattern recognition were proposed recently, namely the microsleep (MS) and the alpha burst (AB) pattern recognition. We compare both methods by applying them to the same experimental data and relating their output variables to two independent variables of driver drowsiness. The latter was an objective lane tracking performance variable and the first was a subjective variable of self-experienced sleepiness. Results offer remarkable differences between both EEG analysis methodologies. The expected increase with time since sleep as well as with time on task, which also exhibited in both independent variables, was not identified after applying AB recognition. The EEG immediately before fatigue related crashes contained both patterns. MS patterns were remarkably more frequent before crashes; almost every crash (98.5 %) was preceded by MS patterns, whereas less than 64 % of all crashes had AB patterns within a 10 sec pre-crash interval.

Current Directions in Biomedical Engineering, 2019

This paper examines the question of how strongly the spectral properties of the EEG during microsleep differ between individuals. For this purpose, 3859 microsleep examples were compared with 4044 counterexamples in which drivers were very drowsy but were able to perform the driving task. Two types of signal features were compared: logarithmic power spectral densities and entropy measures of wavelets coefficient series. Discriminant analyses were performed with the following machine learning methods: support-vector machines, gradient boosting, learning vector quantization. To the best of our knowledge, this is the first time that results of the leave-one-subject-out cross-validation (LOSO CV) for the detection of microsleep are presented. Error rates lower than 5.0 % resulted in 17 subjects and lower than 13 % in another 11 subjects. In 3 individuals, EEG features could not be explained by the pool of EEG features of all other individuals; for them, detection errors were 15.1 %, 17....

43rd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), 2021

A microsleep (MS) is a complete lapse of responsiveness due to an episode of brief sleep (≲ 15 s) with eyes partially or completely closed. MSs are highly correlated with the risk of car accidents, severe injuries, and death. To investigate EEG changes during MSs, we used a 2D continuous visuomotor tracking (CVT) task and eye-video to identify MSs in 20 subjects performing the 50-min task. Following preprocessing, FFT spectral analysis was used to calculate the activity in the EEG delta, theta, alpha, beta, and gamma bands, followed by eLORETA for source reconstruction. A group statistical analysis was performed to compare the change in activity over EEG bands of an MS to its baseline. After correction for multiple comparisons, we found maximum increases in delta, theta, and alpha activities over the frontal lobe, and beta over the parietal and occipital lobes. There were no significant changes in the gamma band, and no significant decreases in any band. Our results are in agreement with previous studies which reported increased alpha activity in MSs. However, this is the first study to have reported increased beta activity during MSs, which, due to the usual association of beta activity with wakefulness, was unexpected.

Current Directions in Biomedical Engineering, 2016

This contribution addresses the question if imminent changes of the cortical state are predictable. The analysis is based on 1484 examples of microsleep (MS) and 1940 counterexamples of sustained attention (SA), both observed during overnight driving in the simulator. EEG segments (8 s in length) immediately before each respective event were included. Features were extracted by (i) modified periodogram and (ii) Choi-Williams distribution. Machine learning algorithms, namely the optimized learning vector quantization (OLVQ) and the support-vector machine with Gaussian kernel function (SVM), were trained in order to map signal features to the event type (MS or SA). Cross validation analysis yielded test set classification accuracies of 87.5 ± 0.1 % and 82.7 ± 0.1 % for feature set (i) and (ii), respectively. In general, SVM outperformed OLVQ. In conclusion, EEG contains enough information to predict immediately upcoming microsleep events.

2019

A microsleep is a brief lapse in performance due to an involuntary sleep-related loss of consciousness. These episodes are of particular importance in occupations requiring extended unimpaired visuomotor performance, such as driving. Detection and even prediction of microsleeps has the potential to prevent catastrophic events and fatal accidents. In this study, we examined detection and prediction of microsleeps using EEG data of 8 subjects who performed two 1-h sessions of continuous 1-D tracking. A regularized spatio-temporal filtering and classification (RSTFC) method was used to extract features from 5-s EEG segments. These features were then used to train three different linear classifiers: linear discriminant analysis (LDA), sparse Bayesian learning (SBL), and variational Bayesian logistic regression (VBLR). The performance of microsleep state detection and prediction was evaluated using leave-one-subject-out cross-validation. The detection performance measures were AUCROC 0.9...

An overview of several methods of electroencephalography (EEG) analysis in order to assess driver sleepiness is presented. All methods were applied to one single data set obtained from overnight driving simulations in our lab. 10 young adults (age 22.4 ± 4.1 years) participated and drove on rural roads; time on task was 7 x 40 min and time since sleep ranged between 16 and 22 hours. Results show large inter-individual variability of all variables and moderate correlation coefficients to one subjective and one objective independent variable of driver drowsiness. Only one method, the detection of microsleep-like EEG patterns, provides a variable with strong increases immediately before sleepiness related crashes. It is concluded that EEG analysis should attach more importance to shortterm patterns and should renounce the analysis of spectral power in four bands.

An overview of several methods of electroencephalography (EEG) analysis in order to assess driver sleepiness is presented. All methods were applied to one single data set obtained from overnight driving simulations in our lab. 10 young adults (age 22.4 ± 4.1 years) participated and drove on rural roads; time on task was 7 x 40 min and time since sleep ranged between 16 and 22 hours. Results show large inter-individual variability of all variables and moderate correlation coefficients to one subjective and one objective independent variable of driver drowsiness. Only one method, the detection of microsleep-like EEG patterns, provides a variable with strong increases immediately before sleepiness related crashes. It is concluded that EEG analysis should attach more importance to shortterm patterns and should renounce the analysis of spectral power in four bands.

2007

An adaptive biosignal analysis method for the detection and extraction of microsleep events is presented. We proposed a blind source extraction method applying Kalman filtering to extract the microsleep events. This is achieved by an adaptive algorithm in which the cost function jointly estimates the kurtosis and a measure of nonlinearity. Next, Kalman filtering is applied to blindly extract the signal of interest. This method was applied to the electroencephalogram and electroculogram recorded of 23 young volunteers while performing monotonic overnight driving in a real car driving simulation laboratory. The extracted microsleep signals can then be used for driving simulation pilot studies for alertness monitoring, and to trigger the activation of alertness countermeasures.

2019

A microsleep is a brief and involuntary sleep-related loss of consciousness of up to 15 s during an active and attention-demanding task. Such episodes of unresponsiveness are of particularly high importance in people who perform high-risk and monotonous activities requiring extended- attention and unimpaired visuomotor performance, such as car and truck drivers, train drivers, pilots, and air-traffic controllers, where microsleeps can, and do, result in catastrophic accidents and fatalities. Microsleep-related accidents can potentially be avoided and thereby lives saved, if microsleeps are noninvasively and accurately predicted. The aim of this study was to explore various inter-channel relationships in the electroencephalogram (EEG) for detection/prediction of microsleeps. In addition to feature-level and decision-level data fusion techniques, ensemble classification techniques were investigated to improve microsleep detection/prediction accuracies. The data used in this research w...