Automated detection of congestive heart failure based on the eigenvalue decomposition of HRV signals

Abstract

Electrocardiogram (ECG) is a noninvasive diagnostic tool which is widely used to diagnose the cardiovascular diseases (CVD). Heart rate variability (HRV) signals are extracted from ECG. It contains the relevant information of the cardiac movements. Congestive heart failure (CHF) is a cardiac disease in which heart is not able to pump sufficient blood to all the parts of the body. This study aims to diagnose the CHF accurately using HRV signals. We have used eigenvalue decomposition of Hankel matrix (EVDHM) method to decompose the HRV signals. The criteria to select the significant decomposed components are defined in this work. Thereafter, nine features corresponding to the five parameters: mean and standard deviation of the signal, mean frequency calculation using Fourier Bessel series expansion, k-nearest neighbour ( kNN) entropy, and correntropy are evaluated from the decomposed components. The obtained features obtained are normalized with z-score normalisation method and then the student’s t-test is used to evaluate the differentiation ability of the features. The ranked features based on t-values are then supplied as input to the least-squares support vector machine (LS-SVM) classifier with radial basis function (RBF) kernel for automated diagnosis of CHF HRV signals. We have tested our method for three combinations of dataset. The combination with the best results obtained an accuracy of 98.50%, sensitivity of 97.80%, and specificity of 99.20% with HRV signals of the segment length of 500 samples and an accuracy of 98.83%, sensitivity of 99.23% and specificity of 98.33% for HRV signals corresponding to the segment length of 2000 samples. Our proposed method can aid the cardiac physicians in accurate diagnosis of CHF patients. Hence, it will help in providing timely treatment to CHF patients.