Hybrid Approach of EEG Stress Level Classification Using K-Means Clustering and Support Vector Machine

Support vector machine (SVM) algorithms are prevalent in classifying electroencephalogram (EEG) signals for the detection of mental stress at various levels. This study aimed to reduce the subjective bias in form of human stress reactivity, by employing clustering methods to pre-label stress levels according to the inherent homogeneity and, perform SVM to classify the stress level. Brainwave signals at the prefrontal cortex (Fp1 and Fp2) from 50 participants were captured related to the stress induced by the virtual reality (VR) horror video and intelligence quotient (IQ) test. The power spectral density (PSD) values of Theta, Alpha, and Beta frequency bands were extracted, and Wilcoxon signed-rank test were reported to show a significant difference in the absolute power between resting baseline and post-stimuli. The extracted features were further clustered into three groups of stress level. The labelled data based on k-means clustering method were fed into SVM to classify the stress levels. The performance of SVM classifier was validated by 10-fold cross validation method and the result affirmed the highest performance of 98% accuracy by using only the feature of Beta-band absolute power at right (Fp2) prefrontal region on account of the significant changes of Beta activity during pre- and post-stimuli. In essence, stress pattern has been found in the brain activity of Beta frequency band within right prefrontal cortex which has been shown to be significantly more active under stimuli. The hybrid approach of classification using k-means clustering and SVM has been proven to be an effective method in lieu of pre-labelling the stress level to reduce individual differences in stress response, and in turn to improve the reliability and detection rate of mental stress.