Normalised corrected entropy analysis of heartbeat dynamics in congestive heart failure and arrhythmia patients

It is a well-known fact that studying entropies in physiological systems, like the cardiovascular system, to quantify dynamic complexity is a well-established field that aids in autonomous health condition evaluation, reducing the need for continuous expert monitoring. Pathological cardiac conditions such as Congestive Heart Failure (CHF) patients and Atrial Fibrillation (AFib) often pose a risk of co-occurrence, which can be lethal. The primary goal of this study is to collectively analyse the symbolic dynamics of human heart rhythms obtained from normal (healthy), CHF, and AFib subjects, using Normalised Corrected Shannon Entropy (NCSE). This study performs joint differentiation of CHF and AFib conditions, instead of performing individual detection as done in previous studies. Initially, the RR time series data, both filtered and raw, from ECGs of 15 subjects are symbolised and converted into series of codes. These code series are further evaluated to compute absolute and mean NCSE values across different thresholds and word lengths. In all the scenarios, the NCSE outcomes showed significant differences between normal and diseased cardiovascular systems, with AFib exhibiting the highest entropy, followed by normal and CHF systems. The results of NCSE and Shannon entropy (SE) were compared and it was observed that NCSE demonstrated superior performance in separating the three conditions, as well as in executing joint detection. The NCSE results have also been statistically analysed and validated using surrogate data analysis, one-way ANOVA and pairwise <i>t</i>-tests. Furthermore, this approach did not require any classification algorithms or noise cancellation methods, indicating robustness to noise.