HRV regularity during persistent atrial fibrillation: A parametric assessment using sample entropy

In this study, we investigated the relation between sample entropy (SampEn) of HRV series and the connected theoretical value (SE_TH), obtained for the autoregressive (AR) models fitted to the same sequences. AR models are commonly used for parametrical spectral analysis and classical HRV spectral parameters were considered as well. The analysis was performed on a subpopulation of the Rate Control in Atrial Fibrillation (RATAF) study, where RR series were collected before and after a β-blocker, Carvedilol, was administered. SampEn, SE_TH and the spectral parameters were significantly different after drug administration. However while SampEn is sensible to nonlinearities or non-Gaussianity in the series, the other parameters are not. To investigate further the changes in the series induced by the drug, both synthetic series generated by the fitted AR models and IAAFT surrogates were employed. The results suggest a reduction in non-Gaussianity as long as a relatively smaller increase in regularity.