Why the study?
The study aimed to develop an objective stress assessment framework by investigating the correlation between EEG signals and Perceived Stress Scale scores using data segmentation and machine learning techniques.
Machine learning algorithms, particularly AdaBoost and Random Forest, can accurately classify perceived stress levels into up to four categories using short segments of EEG data.
Answers are grounded in this paper and cite their sources.
EEG-ML models achieve high multi-class stress accuracy; hypothesis-generating and requires larger prospective validation before clinical use.
This paper focuses on utilizing Electroencephalography (EEG) signals and machine learning techniques in developing an objective stress assessment framework. The study aimed to investigate the correlation between EEG and Perceived Stress Scale (PSS) by utilizing data segmentation technique. The PSS scores are employed to record perceived stress levels of individuals. These PSS scores serve as the basis for categorizing the data into three classes: i) two class: stressed and non-stressed ii) three class: stressed, mildly stressed, and non-stressed, iii) four class: highly stressed, moderately stressed, mildly stressed and non-stressed. EEG recordings are captured from 40 participants using 4 channels Inter axon Muse headband, equipped with dry electrodes. The EEG data is segmented into units of 10 seconds. The data is processed to extract five feature sets including Power Spectrum, Rational Asymmetry, Differential Asymmetry, Correlation and Power Spectral Density. The success levels are accessed utilizing classifiers (Naive Bayes, Support Vector Machine, Logistic Regression, Simple Logistic Regression, Random Tree, K-Nearest Neighbor, Bagging, Random Forest, Multilayer Perceptron, AdaBoost). The highest accuracies achieved for two-, three-, and four-class stress classification are 91.52%, 88.47%, and 87.36%, respectively. These accuracies are obtained using the Adaboost classifier for two-class classification, the Random Forest classifier for three-class classification, and the Adaboost classifier again for four-class classification. These findings underline the importance of the chosen features and classifiers in increasing the prediction accuracy while contributing to the existing knowledge on stress detection with EEG Signals.
No takes yet. Share an insight, caveat, or question.
Buzdar et al. (2023) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: