Automatic epileptic seizure detection in EEGs based on optimized sample entropy and extreme learning machine

► We propose a novel optimized sample entropy algorithm for epileptic seizure detection. ► Sample entropy with high statistical significance represents the characteristics of epileptic EEGs. ► Optimal parameters of sample entropy are found for epileptic EEG signal analysis. ► A sample entropy-extreme learning machine framework is proposed for detecting epileptic seizures in EEGs. ► The proposed method achieves not only a high detection accuracy but also a very fast computation speed.