Electroencephalography as a predictor of self-report fatigue/sleepiness during monotonous driving in train drivers

Abstract

Objective: In this study, electroencephalography activity recorded during monotonous driving was investigated to examine the predictive capability of monopolar EEG analysis for fatigue/sleepiness in a cohort of train drivers. Approach: Sixty-three train drivers participated in the study, where 32- lead monopolar EEG data was recorded during a monotonous driving task. Participant sleepiness was assessed using the Pittsburgh sleep quality index (PSQI), the Epworth sleepiness scale (ESS), the Karolinksa sleepiness scale (KSS) and the checklist of individual strength 20 (CIS20). Main results: Self-reported fatigue/sleepiness scores of the train driver cohort were primarily associated with EEG delta, theta, and alpha variables; however, some beta and gamma associations were also implicated. Furthermore, general linear models informed by these EEG variables were able to predict self-reported scores with varying degrees of success, representing between 48% and 54% of variance in fatigue scores. Significance: Self-reported fatigue/sleepiness scores of train drivers were predicted with varying degrees of success (dependent upon the self-reported fatigue/sleepiness measure) by alterations to monopolar delta, theta, and alpha band activity variables, indicating EEG as a potential indicator for fatigue/sleepiness in train drivers.

Human resting-state EEG and radiofrequency GSM mobile phone exposure: the impact of the individual alpha frequency

“…For example, several external factors could affect the alpha band power spectral density, such as nicotine (Teneggi et al 2004;Domino et al 2009), alcohol (Lithari et al 2012;Rosen et al 2014) and caffeine consumption (Siepmann and Kirch 2002;Deslandes et al 2005) or emotional stress behavior (Lewis et al 2007;Al-Shargie et al 2016) and physiological hormonal variation, such as melatonin (Cajochen et al 1998), cortisol (Sannita et al 1999;Tops et al 2005) and female ovarian hormones (Becker et al 1982;Brötzner et al 2014). In addition, the sleep-wake cycle could determine the level of alertness, modulating the waking EEG and the alpha band amplitude (Cajochen et al 1995;Lafrance and Dumont 2000;Regen et al 2013;Lees et al 2018). Some of previous studies which investigated the RF-EMF effect on the waking EEG considered abovementioned factors, in order to limit potential interaction between these cofounding factors and EEG modulation (for review see (Danker-Hopfe et al 2019;Wallace and Selmaoui 2019).…”

Machine learning polysomnographically-derived electroencephalography biomarkers predictive of epworth sleepiness scale

“…Several approaches have been investigated to develop an objective neurophysiologic biomarker capable of capturing symptoms of EDS. For example, the least absolute shrinkage and selection operator (LASSO) was used to predict ESS from EEG signals collected from train drivers, but with varying degrees of success and requiring more complex computational techniques compared to our study, which was guided by a statistical approach for the selection of ML features 18 . Another proposed sleepiness biomarker is the odds ratio product, computed from the delta, theta, alpha-sigma, and beta frequency bands from EEG signals, and its association to ESS 19 .…”

Effect of Split Condition and Driving Duration on Fatigue Based on Changes in Brain Wave Signals when Driving a Train Simulator

“…The EEG used in this study was the EPOC+ EEG (Emotiv Inc., USA), which uses the standard 10-20 electrode pairing. The lobes reviewed in this study were the frontal and occipital lobes, which according to various studies can detect fatigue well [11].…”

“…The experiment was then closed by filling out the SOFI questionnaire (15 minutes), followed by the measurement of tension and body temperature, Visualization of simulation time can be seen in figure The collection data in this study are changes in EEG signal waveforms and SOFI values at the beginning and end of the simulation. EEG signal waveform recording data needs to be processed through filtration and transformation first [11]. The results of pre-processing EEG data will then be further transformed into power spectral density (PSD), which is the average value of each brain wave.…”

“…Koudelková, Strmiska, and Jašek (2018) divided delta waves at frequencies of 0.5-4 Hz, teta at frequencies of 4-8 Hz, alpha at frequencies of 8-13 Hz, beta at frequencies of 13-30 Hz, and gamma at frequencies greater than 30 Hz. Research related to fatigue in the context of driving by Lees et al [11] showed an increase in wave activity at low frequencies, namely delta, teta, and alpha, and a decrease in waves at high frequencies, namely beta and gamma.…”