CLASSIFICATION OF COGNITIVE STRESS AS A PSYCHOLOGICAL INDICATOR THROUGH MACHINE LEARNING

In this study, we explored the potential of Support Vector Machine (SVM) method for classifying levels of cognitive stress using EEG (Electroencephalogram) signals. The goal is to develop accurate models that would enable the prediction and understanding of not only the current mental state of the subjects, but also potential real-time interventions. In medical fields, the application can be seen in the treatment of attention, focus, hyperactivity, autism, and depression disorders. Additionally, there is an extremely high potential for application in areas such as psychology, sociology, education, economics, neuromarketing, security, and in enhancing workplace stress management, anxiety treatment, digital marketing, economicfinancial forensics, as well as improving user experience in virtual environments and video games. The results have shown that it is possible to differentiate high and low levels of cognitive stress with satisfactory accuracy, opening the way for the application of these findings in various fields. Cognitive stress represents one of the fundamental cognitive processes that causes individuals to behave and think differently in certain situations than in their usual state of consciousness. Predicting, analyzing, and understanding the level of cognitive stress from EEG signals is of great importance in various fields, including neuroscience, psychology, education, professional sports, human-computer interaction, and many other areas. Machine learning represents a subgroup of artificial intelligence that uses statistical models, and functions to ‘learn’ and ‘train’ data resulting in corresponding output values. The brain-computer interface, through which data on cognitive stress, among other parameters and psychological categories, is collected, is based on the functioning of EEG devices. The prediction of cognitive stress represents the application of machine learning, recording and using brain EEG signals or extracted characteristics from EEG signals as input values, in order to predict the level of output values of cognitive stress, of high or low degree, reflecting the mental state of the subjects in real time