Abstract
Prolonged wakefulness is known to adversely affect basic cognitive abilities such as object recognition and decision-making. It affects the dynamics of neuronal networks in the brain and can even lead to hallucinations and epileptic seizures. In cognitive-intensive workplaces, there is a requirement to refine an objective method of quantifying the current level of cognitive capabilities, rather than relying on subjective self-reporting. In this study, we compiled electroencephalography (EEG) recordings from several sleep-deprivation workshops held by the Israeli Air Force, done by flight cadets and unmanned aerial vehicle operators. By extracting a wide range of EEG features and applying machine-learning techniques, we were able to accurately predict the reaction time of participants undergoing a simple psychomotor vigilance task, which acted as a stand-in for basic cognitive functions. Furthermore, through the use of interpretability methods, we examined the importance of different EEG features and their contribution to changes in the behavioral metrics.