Stress Monitoring in Humans using Biomedical Signal Analysis

Nowadays, stress is one of the major issues in every individual's life and may cause both physiological and psychological problems. Researchers have taken this into account and proposed various stress detection models, which can be measured by using biomedical signals like Electrocardiogram (ECG), Electromyogram (EMG), Electroencephalogram (EEG) and Electroneurogram (ENG). EEG signals are capable of measuring and monitoring neurological electrical activity and detecting changes in the brain. The EEG data is gathered from PhysioNet and pre-processed by filtering out noise using a fourth-order Butterworth filter and decomposing it into different frequency bands (y (>30Hz), β (13-30Hz),
$a$
(S-13Hz), θ (4-8Hz), and δ (0.5-4Hz)) using Discrete Wavelet Transform (DWT) from the fifth-order Daubechies wavelet family. A variety of EEG signal characteristics were then extracted to examine EEG signal patterns under different circumstances. Support Vector Machine (SVM) classifier is used in the study to classify the results and aims to improve accuracy and efficiency of stress detection compared to existing models.