Memory beyond memory in heart beating: an efficient way to detect pathological conditions

Abstract

We study the long-range correlations of heartbeat fluctuations with the method of diffusion entropy. We show that this method of analysis yields a scaling parameter $\delta$ that apparently conflicts with the direct evaluation of the distribution of times of sojourn in states with a given heartbeat frequency. The strength of the memory responsible for this discrepancy is given by a parameter $\epsilon^{2}$, which is derived from real data. The distribution of patients in the ($\delta$, $\epsilon^{2}$)-plane yields a neat separation of the healthy from the congestive heart failure subjects.