Complexity analysis of heart beat time series by threshold based symbolic entropy

Abstract

Complex variations have been observed in heart rate. Analysis of these variations, i.e., heart rate variability (HRV) analysis has become an important non-invasive technique to study the sympathovagal interactions in physiological and pathological conditions. Increasing efforts were made in the development of HRV measures for quantifying heart rate variations in order to make clinically useful assessments of patient welfare. Heart is not a periodic oscillator under normal physiologic conditions and standard linear HRV measures may not be able to detect subtle, but important changes in heart rate time series, whereas, most of nonlinear measures suffer from the curse of dimensionality. To overcome these difficulties, several complexity measures, especially from symbolic dynamics have been proposed. Recently, we have used threshold dependent symbolic entropy to study the dynamics of stride interval time series of control (healthy) and neurodegenerative diseased subjects. Normalized corrected Shannon entropy (NCSE) was used to quantify these dynamics. In this paper, using this technique, we have compared the complexity of normal sinus rhythm (NSR), congestive heart failure (CHF) and atrial fibrillation (AF) subjects. We investigated that the dynamics of healthy (NSR) subjects are more complex than diseased (AF and CHF) subjects within the short range of thresholds.