Detection of Mental Stress using EEG signals - Alpha, Beta, Theta, and Gamma Bands
Savita Bakare, Shruti Kuge, Siri Sugandhi, Shashank Warad, Vinay Panguddi
Abstract
Mental stress, or psychological stress, arises when individuals perceive emotional or psychological strain beyond their coping abilities. Various factors such as personal relationships, work pressure, financial problems, or major life changes, impact both emotional and physical well-being. Electroencephalography (EEG) signals serve as insightful indicators of brain activity, resembling tiny electrical whispers exchanged between brain cells. These brain cells are characterized by EEG patterns such as alpha, beta, theta, and gamma frequency band waves, and offer valuable insights such as frequency analysis, categorization of valence and arousal, and machine learning techniques, into distinct mental states and neural activity. The exploratory data analytics (EDA) techniques using ML methods (KNN, SVM, and RF) on EEG dataset is being performed to analyze mental stress detection. Furthermore, Mean Absolute Error (MAE) scores of the KNN, SVM, and RF models are compared based on the number of folds, which shows that KNN has the lowest MAE of 0.62, compared to SVM scores of 0.83 and RF score of 1.47 respectively. As a result, the KNN model is selected for stress detection due to its superior performance in reducing MAE in both iterations and training data sizes. Furthermore, the study concisely also reviews an existing literature on mental stress detection using EEG signals, highlighting prevalent challenges and research gaps.