<![CDATA[Stress is considered one of the major global health issues contributing to cardiovascular disease and depression among other disorders. This study examines the amounts of stress and performance on tasks using electroencephalogram (EEG) data and machine learning. Raw EEG data is preprocessed to remove noise and segment epochs. Empirical Mode Decomposition (EMD) is followed by Butterfly Optimization Algorithm (BOA) for feature extraction and dimensionality reduction. Five machine learning classifiers (SVM, Naive Bayes, Random Forest, KNN, Decision Tree) classify four levels of stress (neutral, low, medium, and high) based on cognitive load during two tasks: the Stroop color-word task and an arithmetic task. Results indicate the Naive Bayes classifiers for the Stroop and arithmetic tasks had accuracies of 98.82% and 98.87% respectively, while the SVM classifier achieved 99.02% accuracy for both tasks combined. Such results attest to the growing interest and application of machine learning on EEG data for mental health monitoring and the possible enhancement of task performance. .]]>
Copyrights © 2025
