Stochastic modeling and prediction of experimental seizures in Sprague-Dawley rats

Most seizure prediction methods are based on nonlinear dynamic techniques, which are highly computationally expensive, thus limiting their clinical us efulness. The authors propose a different approach for prediction that uses a stochastic Markov chain model. Seizure (T-s) and interictal (T-i) durati ons were measured from 11 rats treated with 3-mercaptopropionic acid. The d uration of a seizure T-s was used to predict the time (T-i2) to the next on e. T-s and T-i were distributed bimodally into short (S) and long (L), gene rating four probable transitions: S --> S, S --> L, L --> S, and L --> L. T he joint probability density f (T-s, T-i2) was modeled, and was used to pre dict T-i2 given T-s. An identical model predicted T-s given the duration T- i1, of the preceding interictal interval. The median prediction error was 3 .0 +/- 3.5 seconds for T-s (given T-i1) and 6.5 +/- 2.0 seconds for T-i2 (g iven T-s). In comparison, ranges for observed values were 2.3 seconds < T-s < 120 seconds and 6.6 seconds < T-i < 782 seconds. These results suggest t hat stochastic models are potentially useful tools for the prediction of se izures. Further investigation of the probable temporal interdependence betw een the ictal and interictal states may provide valuable insight into the d ynamics of the epileptic brain.