Seizure anticipation in pediatric epilepsy: use of Kolmogorov entropy.

The purpose of this paper is to demonstrate feasibility of using trends in Kolmogorov entropy to anticipate seizures in pediatric patients with intractable epilepsy. Surface and intracranial recordings of preseizure and seizure activity were obtained from five patients and subjected to time series analysis using Kolmogorov entropy. This metric was compared with correlation dimension and power indices, both known to predict seizures in some adult patients. We used alarm levels and introduced regression analysis as a quantitative approach to the analysis of trends. Surrogate time series evaluated data nonlinearity, as a precondition to the use of nonlinear measures. Seizures were anticipated before clinical or electrographic seizure onset for three of the five patients from the intracranial recordings, and in two of five patients from the scalp recordings. Anticipation times varied between 2 and 40 minutes. This is the first report in which simultaneous surface and intracranial recording are used for seizure prediction in children. We conclude that the Kolmogorov entropy and power indices were as effective as the more commonly used correlation dimension in anticipating seizures. Further, regression analysis of the Kolmogorov entropy time series is feasible, making the analysis of data trends more objective.