Comment on: “Performance of a seizure warning algorithm based on the dynamics of intracranial EEG”

Abstract

With great interest we read the article ofChaovalitwongse et al. (2005) concerning the per-formance of an automated seizure warning system(ASWS) based on concepts from nonlinear dynam-ics. To assess the performance of their algorithm, theauthors divided long-term intracranial EEG data from10patientsintotrainingandtestdatasets.Forthetrain-ing data, the authors reported a high prediction perfor-mance with an average sensitivity of 76.12% acceptingan average false prediction rate of 0.17 false warningsper hour. For the test data, an average sensitivity of68.75% and an average false prediction rate of 0.15were obtained.These promising results nurture the hope to estab-lish a therapeutic device for epilepsy patients basedon an in-time seizure warning. However, seizure pre-diction suffers from the intrinsic problem that high