Deterministic nonlinearity in ventricular fibrillation

We provide numerical evidence that the electrocardiogram data collected fro m pigs during induced ventricular fibrillation cannot be described by a mon otonic nonlinear transformation of linearly filtered noise. To establish th is we use surrogate techniques and apply two test statistics: (1) the Taken s' maximum likelihood estimator of the Grassberger-Procaccia correlation di mension and (2) an improved correlation dimension estimation routine. The i mproved dimension estimates provide evidence that the correlation dimension of the underlying dynamics during the episode of VF in the first 30 s is s lightly less than 6. This result is consistent and reproducible among subje cts. (C) 2000 American Institute of Physics. [S1054-1500(00)00601-7].