Comparative Analysis of Heart Rate Variability Signals

Abstract

Heart Rate Variability Analysis is the method of evaluating the state of mechanisms for regulating the physiological functions in the human beings and animals. It is a measure of neuro-cardiac function reflecting the autonomic nervous system and heart-brain interactions. The study and investigation of HRV mainly have centre of interest on the process of analysis of fluctuations in inter-beat intervals in the heart rate and the ability of diagnosis provided by the fluctuations. The information obtained from the records of HRV has eminent importance for the clinicians and researchers for the identification of the nature of any illness or symptoms. The main focus of this dissertation work is the implementation of the visibility graph method and calculation of network measures which can be obtained from the network graph constructed from RR interval segments over long term Electrocardiogram (ECG) recordings. Time domain, frequency domain and nonlinear methodologies for HRV analysis has been studied and reviewed. We evaluated mainly three network measures such as characteristic path length (CPL), average clustering coefficient (ACC), and standard deviation of the shortest path length (CPL-STD) of the network graph obtained from the RR interval segment on scale of different data points for Congestive Heart Failure (CHF) and Normal Sinus Rhythm (NSR) signals. Time domain, frequency domain and Poincare plot measures for HRV was analyzed using Kubios HRV analysis software. All the network measures estimated from RR interval segment showed statistically significant result between Normal Sinus Rhythm and Congestive Heart Failure subjects on different scale of data points except the Characteristic Path Length (CPL) on scale of 1500 data points and CPL, and standard deviation of shortest path length (CPL-STD) on 2000 data points. Characteristic Path Length (CPL) was found to be more in Normal Sinus Rhythm subjects and decreased in Congestive Heart Failure subjects while Average Clustering Coefficient has increased value in case of Congestive Heart Failure patients. Statistically significant time domain measures were found to be lower in CHF as compared to NSR. Frequency domain measure LF/HF ratio was lower in CHF. For nonlinear analysis using Poincare plot showed lower values of SD1 and SD2 in CHF and higher value of SD1 and SD2 in NSR while SD2/SD1 ratio was found to be higher in NSR. Time domain analysis results have shown an increase of HR and reduction of HRV in CHF as compared to NSR and hence, a higher heart rate in combination of lower Heart Rate Variability is a well known indicator of (acute) stress.