Mental Task Induced Stress Detection using Multi-Variate Weighted Visibility Graph (MV-WVG) from EEG Signals

The detection of stress induced by mental tasks is crucial for understanding the impact of cognitive activities on an individual’s well-being. Electroencephalogram (EEG) signals hold significant promise as a non-invasive and objective method for assessing cognitive states. In this study, we propose an innovative approach for simultaneously detecting mental activity and task-induced stress from EEG signals by learning multi-variate weighted visibility graphs with Graph Signal Processing (GSP) techniques. Spectral graph features that are indicative of stress-related patterns are extracted from the graph, and the benchmark classifiers are then trained on these features to detect the stressed state from mental tasks. To evaluate the proposed method’s efficacy, experiments are conducted on two different datasets comprising EEG recordings from individuals performing various cognitive tasks, ranging from simple to complex, under controlled stress-inducing conditions. The results showcase the effectiveness of the suggested method in accurately distinguishing various mental activities and the stress they induce.