EEG based Stress Analysis through Feature Extraction

Abstract

The diagnosis of Stress relies virtually solely on doctor-patient conversation and scale analysis, which includes problems such as patient denial, insensitivity, subjective biases, and inaccuracy. Improving the accuracy of Stress diagnosis and therapy necessitates the development of an objective, computerized system for predicting clinical outcomes. Using the modification of EEG data and machine learning techniques, this study attempts to improve the recognition of Stress. The EEG data of 10 volunteers were acquired using a Narosky device during an experiment, including emotive facial stimuli. Psychiatrists used the EEG signal as the criterion for diagnosis of Stress in patients. The different approaches processed the features: machine learning and deep learning. Significant outcomes are achieved using PCA, ICA & EMD for BCI applications. SVM empowers a developer with several advantages: PCA exhibits excellent generalization properties, with stress & pressure detection using EEG Signals. If the signals are negative, the impact of overtraining is sensitive to the curse-of- dimensionality. These advantages were achieved by using EEG signals to detect Stress. The experimental analysis gives some overview of all different approaches, which depend on frequency domain analysis with 14 fourteen-channel EEG signals with reasonable accuracy.