Accurate automated detection of congestive heart failure using eigenvalue decomposition based features extracted from HRV signals

Congestive heart failure (CHF) is a cardiac abnormality in which heart is not able to pump sufficient blood to meet the requirement of all the parts of the body. This study aims to diagnose the CHF accurately using heart rate variability (HRV) signals. The HRV signals are non-stationary and nonlinear in nature. We have used eigenvalue decomposition of Hankel matrix (EVDHM) method to analyze the HRV signals. The lowest frequency component (LFC) and the highest frequency component (HFC) are extracted from the eigenvalue decomposed components of HRV signals. After that, the mean and standard deviation in time domain, mean frequency calculated from Fourier-Bessel series expansion, k-nearest neighbor (k-NN) entropy, and correntropy features are evaluated from the decomposed components. The ranked features based on t-value are fed to least-squares support vector machine (LS-SVM) classifier with radial basis function (RBF) kernel for automated diagnosis of CHF HRV signals. The study is performed on three normal datasets and two CHF datasets. Our proposed system has yielded an accuracy of 93.33%, sensitivity of 91.41%, and specificity of 94.90% using 500 HRV samples. The automated toolkit can aid cardiac physicians in the accurate diagnosis of CHF patients to confirm their findings with our system. Hence, it will help to provide timely treatment for CHF patients and save life. (C) 2018 Nalecz Institute of Biocybernetics and Biomedical Engineering of the Polish Academy of Sciences. Published by Elsevier B.V. All rights reserved.