Electroencephalogram Feature Extraction Methods for Stress Analysis

When diagnosing stress, relying entirely on doctor-patient conversation and scale analysis can lead to problems such as patient denial, insensitivity, subjective biases, and inaccuracy. Improving the accuracy of stress diagnosis and therapy demands the construction of an objective, automated system for predicting clinical outcomes. The goal of this study is to improve stress detection by modifying electroencephalogram (EEG) data and applying machine learning (ML) techniques. EEG data were collected from ten participants using a NeuroSky device during the presentation of different facial expressions. The EEG measurement was used by psychologists to diagnose stress in their patients. ML and deep learning (DL) were the methods that handled the feature processing. Significant outcomes can be achieved with brain-computer interface (BCI) applications using principal component analysis (PCA), independent component analysis (ICA), and empirical mode decomposition (EMD). Using support vector machine (SVM) in the software development processes has various advantages: It is possible to utilize principal component analysis (PCA) to detect stress and pressure in EEG readings because of its excellent generalization features. While over-training (intensive preparation) can yield good results in some cases, it may suffer from the curse of dimensionality when the input signals are high-dimensional or poorly structured.