Revisiting sample entropy analysis

We modify the definition of sample entropy (SaEn) by incorporating a time delay between the components of the block (from which the densities are estimated) and show that the modified method characterizes the complexity of the system better than the original version. We apply the modified SaEn to the standard deterministic systems and stochastic processes (uncorrelated and long range correlated (LRC) processes) and show that the underlying complexity of the system is better quantified by the modified method. We extend this analysis to the RR intervals of the normal and congestive heart failure (CHF) subjects (available via www.physionet.org) and show that there is a good degree of separation between the two groups.