Analysis of Normal and Pathological Heart Rate Variability Based on Electrocardiogram Data

Abstract

The heart rate variability analysis is carried out using mathematical methods in the time domain, frequency domain and nonlinear methods. The electrocardiographic records in normal and cardiac pathology from the open research resource PhysioNet were materials of the study. A database of the results of the various patient groups analysis was formed. A comparative analysis of the indicators revealed statistically significant differences in most variability indicators between normal rhythm patient groups. patient groups with class I CHF and patient groups with II, III CHF classes. The LASSO method revealed the main, most significant indicators can be used to fully characterize of the rhythm variability, as well as the possible detection its normal or pathology. Based on these indicators, patient clustering was carried out in order to distinguish two groups: the normal and the cardiac pathology, while the quality of the clustering was assessed by the external metric (the Rand index).