Stress Detection Based on ECG Using Discrete Wavelet Transform

Abstract

Acute stress is the most common form of stress. It comes from demands and pressures of the recent past and anticipated demands and pressures of the near future. This research studied the stress on female students due to mathematical exercises in a noisy environment. Detection of this stress is important because it contributes to diverse pathophysiological changes including sudden death, ischemic diseases (myocardial infarction, angina), and wall motion abnormalities (the motion of a region of the heart muscle is abnormal), as well as to alterations in cardiac regulation as indexed by changes in sympathetic nervous system activity and hemostasis (process which causes bleeding to stop in order to keep blood within a damaged blood vessel unlike hemorrhage). Stress level is difficult to manage because it cannot be measured in a consistent and timely way. One current method to characterize an individual’s stress level is to conduct an interview or to administer a questionnaire during a visit with a physician or psychologist. HRV (Heart rate variability) can be analyzed using both time domain and frequency domain features. Selection of features which vary with the changes of the stress levels is significant and it is important to show relatively reliable behavior. Overall, heart rate variability spectra during baseline conditions related to Left ventricular hypertrophy and congestive heart failure are dominated by high frequency activity. Stress is accompanied by an increase in the Power Spectrum Density (PSD) of Low Frequency (LF) and decrease in PSD of High Frequency (HF). Data (ECG signal) was collected by AD (Data acquisition) Instrument from ten female subjects, in the age range of 20 to 24 years were of asked to perform three levels difficulties of arithmetic tasks. A total of ten statistical features were used in this research extracted through wavelet transform, including: Mean, Maximum, Minimum, Standard deviation, Variance, Mode, Median, power spectral density (PSD), energy, entropy and hybrid of them. The SVM (support vector machines) classifier give highest accuracy of 79.5 based on hybrid feature and ribo 3.7 wavelet through LF range.

Keywords

Stress, ECG, SVM, KNN, DWT