Extraction of dynamical information and classification of heart rate variability signals using scale based permutation entropy measures

Abstract

The dynamical fluctuations in the Heart Rate Variability (HRV) signals show structures at multiple time scales revealing that complexity of the autonomic nervous system control of the heart is multiscale and hierarchical. Multiscale Entropy (MSE) and its variant Composite MSE (CMSE) were proposed to quantify the complexity at multiple time scales, however, these measures failed to quantify complexity accurately for short duration signals at large temporal scales. To address the downsides of MSE and CMSE, Multiscale Permutation Entropy (MPE) and Improved MPE (IMPE) were proposed. The preliminary results reveal that MPE and IMPE were able to distinguish healthy and pathological subjects, however, further studies are needed to investigate the robustness of these measures. In this study, we investigate the robustness of scale based PE measures in terms of dynamical information, induction of undefined entropy estimates for short duration signals and to classify HRV signals under different physiological and pathological conditions. The results were compared with SE, PE, MSE and CMSE. The MPE and IMPE along with MSE and CMSE provided accurate dynamical information. The results revealed that MPE and IMPE resolved the issue of inducing undefined entropy estimates and are robust in classifying healthy and different pathological subjects.