Discrimination power of long-term heart rate variability measures for chronic heart failure detection

Paolo Melillo,Leandro Pecchia,Mario Sansone,Roberta Fusco,Marcello Bracale

doi:10.1007/s11517-010-0728-5

Abstract

The aim of this study was to investigate the discrimination power of standard long-term heart rate variability (HRV) measures for the diagnosis of chronic heart failure (CHF). The authors performed a retrospective analysis on four public Holter databases, analyzing the data of 72 normal subjects and 44 patients suffering from CHF. To assess the discrimination power of HRV measures, an exhaustive search of all possible combinations of HRV measures was adopted and classifiers based on Classification and Regression Tree (CART) method was developed, which is a non-parametric statistical technique. It was found that the best combination of features is: Total spectral power of all NN intervals up to 0.4 Hz (TOTPWR), square root of the mean of the sum of the squares of differences between adjacent NN intervals (RMSSD) and standard deviation of the averages of NN intervals in all 5-min segments of a 24-h recording (SDANN). The classifiers based on this combination achieved a specificity rate and a sensitivity rate of 100.00 and 89.74%, respectively. The results are comparable with other similar studies, but the method used is particularly valuable because it provides an easy to understand description of classification procedures, in terms of intelligible "if … then …" rules. Finally, the rules obtained by CART are consistent with previous clinical studies.

Highlights

Heart Rate Variability (HRV) is the variation over time of the period between consecutive heartbeats (RR intervals)[30] and is a non-invasive measure commonly used to assess the influence of the autonomic nervous system (ANS) on the heart[23]
The path to the terminal node 2 in the Fig. 2a can be read as: “if TOTPWRLS is higher than 8271.86 ms2 and RMSDD is HRV measures for Chronic Heart Failure (CHF) detection Page nr. 5 of 8 higher than 15.62 ms, the subject is classified as normal”
A final classification split is based on SDANN, that is, if it is lower than 106.71 ms, the subject is classified as CHF patient, otherwise as a normal subject

Summary

Introduction

Heart Rate Variability (HRV) is the variation over time of the period between consecutive heartbeats (RR intervals)[30] and is a non-invasive measure commonly used to assess the influence of the autonomic nervous system (ANS) on the heart[23]. HRV has been widely studied in patients suffering from Chronic Heart Failure (CHF)[1,2,3, 5, 10,11, 13, 17, 19, 27, 32]. Isler et al [13] investigated the discrimination power of short-term HRV measures, including wavelet entropy. In this study, they achieved the best performance using Genetic Algorithms and kNearest Neighbour Classifier, resulting in a sensitivity rate of 100.00% and a specificity rate of 94.74%. This study reached interesting results, the classifier proposed by Isler [13] relied on complex features and rules which are difficult for clinicians to interpret. In both these studies[3, 13] the classifiers were developed using a subset of the dataset adopted in the current study

Objectives

Methods

Results

Conclusion