Defining an instantaneous complexity measure for heartbeat dynamics: The inhomogeneous point-process entropy

Complexity measures have been widely used to characterize the nonlinear nature of cardiovascular control and heartbeat dynamics. Current approaches associate these measures to finite single values within an observation window, thus not being able to characterize instantaneous system dynamics. In this study, we introduce the definition of novel measures of entropy based on the inhomogeneous point-process theory and inspired by the approximate and sample entropy algorithms. The discrete heartbeat series are modeled through probability density functions defined at each moment in time, which characterize and predict the next beat occurrence as a function of the past history through Laguerre expansions of the Wiener-Volterra terms. Experimental results, obtained from the analysis of RR interval series extracted from five ECG recordings during postural and tilt-table maneuvers, suggest that the proposed entropy indices can provide instantaneous tracking of the heartbeat complexity and allow for further definition of the 'complexity variability' framework.