Prediction of epileptic seizure from nonlinear time series of recording

Prediction of epileptic seizure onset, even in a short-term, would provide with time for preventive treatment to minimally keep the risk of seizure and ultimately to improve quality of life. In the present work, the electrocorticogram (ECoG) recordings from Sprague Dawley rat are performed for reconstructing sliding segment firstly, then by means of cross-correlation integral, comparing the ECoG dynamics from sliding segments with a reference epoch (non epileptic). One minute before seizure, the similarity index between the reference and the test epoch approaches to around 1; during the epileptic seizure the index drops to low values and fluctuates. An anticipation time Ta is proposed based on these findings. This technique is helpful to predict epileptic seizure in real practice, and it is a potential and useful strategy in clinical application.