Mental Stress Prediction using Machine Learning on Real Time EEG Signal

Abstract

The primary cause of various health issues is mental stress. Experts and doctors have developed several devices to evaluate the degree of mental strain in the initial phase. The research has suggested several neurological imaging techniques for assessing mental strain at their work. This study uses electroencephalogram (EEG) signals and AI-based techniques to predict mental stress. Real-time EEG data was collected using a brain-computer interface (BCI) strategy with the Mindrove Bright Cap and analyzed using Transform to convert signals into the frequency domain. K-means clustering is used to group the data for stress prediction. Logistic Regression (LR) showed the highest accuracy of 99.77% and an F1-Score of 100% among the models evaluated. Measure the effectiveness of EEG signals and ML in predicting mental stress, with potential applications in identifying stress levels based on brain activity.