Electroencephalogram-Based Stress Index

Stress is one of the major health issues where too much stress may lead to depression, fatigue and insomnia. Various methods have been introduced by researchers to detect and analyze stress level using human physiological signals but yet to come out with a reliable indicator which able to indicate the stress level of healthy human from their brain electrical activity; Electroencephalogram (EEG) signals. This study proposes stress index as an indicator of stress level using EEG signals. The study employs nonparametric method to extract stress features from EEG signals after performing two tasks; do nothing and answer Intelligence Quotient (IQ) test questions. The k-Nearest Neighbor (k-NN) classifier is used to identify the stressed group using the extracted stress features. The results of the study established 3 type of indexes which represent the stress levels (Low Stress, Moderate Stress, High Stress) with 88.89% overall classification accuracy, 86.67% classification sensitivity and 100% classification specificity. The 10-fold and leave-one-out cross validation of the classifier produced classification accuracy of 78.89% and 83.50% respectively.