Nonlinear Modelling of the Daily Heart Rhythm

A model of the Markovian character of the Heart Rate Variability (HRV) is designed by analyzing its information flow. A measure based on higher order cumulants quantifies the dependence of the current value on the past of the time series. That measure is employed as a discriminant statistics to accept or reject the null hypothesis, supposing that a nonlinear Markov process of order n is able to model the given HRV time series. The probability density function characterizing the Markov process is estimated as a sum of Gaussian distributions obtained as outputs of neural networks. The order of the approximating Markov process shows to be a reliable index for quantifying the balance of the autonomic nervous system control activity.