Detection of air traffic controllers’ fatigue using voice analysis - An EEG validation study

Abstract

The article summarizes the results of an EEG validation study following the proposal of a methodology for identification and monitoring of air traffic controllers’ fatigue level based on layered-voice analysis (LVA) described in the initial study (Kouba et al., 2020).10 licensed air traffic controllers from APP Prague participated in the experiment that enabled the comparison of the state of vigilance during certain parts of the shifts rostering cycle. Methods for subjective assessment (Karolinska Sleepiness Scale) of fatigue were used as well as objective methods (Psychomotor Vigilance Task, Oddball Task and EEG power spectra analysis) to compare with the outputs of voice analysis.Our results indicate that the method of voice analysis reacts to changes in a person's mental state. Based on the results of current study, voice parameters marked as Stress and/or Energy levels seem to be the most suitable candidates for fatigue detection as shown by comparative statistics and correlations. The change in these parameters best reflected changes observed in EEG power in different band both during the cognitive testing and during the simulation exercise.According to our research voice analysis is able to identify differences in wakefulness and fatigue, as its results correlate with changes in brain activity. As our experiment likely induced not only fatigue but also changes in other mental states such as perceived stress which could have been reflected in the changes of other voice parameters, this could be the focus of the following studies.