Estimation of Lyapunov exponents of ECG time series—The influence of parameters

The influence of working parameters on the estimation of Lyapunov spectra from limited data sets is investigated, using Lorenz, Rossler and Henon attractors, and an empirical law is presented. Application to electrocardiograms from the MIT-BIH Arrhythmias Database shows a smaller divergence rate (λ1) and convergence rate (¦λ3¦ or ¦λ4¦) in pathological than normal data.