Heart Rate Variability Frequency Domain for Detection of Mental Stress Using Support Vector Machine

Abstract

The Heart Rate Variability (HRV) analysis become a useful tool for understanding the autonomic nervous system (ANS). HRV analysis include spectral analysis or frequency domain defined two major components of the ANS which is interpreted as different physiological rythms. In this work, HRV as a variation of heart rate is used as a tool that describes the activity of the ANS associated with mental stress. The measurements of the HRV frequency domain is used to classify the condition of mental stress and the health condition of subjects related to influenza. The basic frequency domain measurements include nLF, nHF and LF/HF ratio was used as an efficient measurement for stress and influenza classification using SVM Classifier. SVM Classifier are used as a binary classifier for each condition perceived by the subjects. SVM Classifier performance evaluation testing for two different test subjects resulted 94% and 75% accuracy in the classification of influenza from each subject. For stress classification on each subject, we obtained 81% and 66% accuracy using training and testing data from each subject.