Entropy-Based Detection of Complexity and Nonlinearity in Short-Term Heart Period Variability under different Physiopathological States

We compare different estimators of a popular en-tropy-based nonlinear dynamic measure, i.e. the conditional entropy (CE), as regards their ability to assess the complexity and nonlinearity of short-term heart rate variability (HRV). The CE is computed using binning, kernel and nearest neighbor entropy estimators in HRV time series measured from young, old and post-myocardial infarction patients studied at rest and during orthostatic stress. We find that the three estimators yield similar patterns of CE, but different patterns of nonlinear dynamics, across groups and conditions. These results suggest that the strategy for CE estimation is not crucial for the quantification of complexity, but has a remarkable impact on the detection of nonlinear HRV dynamics.