METHOD AND SYSTEM FOR PREDICTING EPILEPTIC SEIZURES BY USING NON-LINEAR TIME SERIES ANALYSES OF PHASE PATTERNS

摘要

PURPOSE: A method and system for predicting epileptic seizures by using non-linear time series analyses of phase patterns are provided to meet effectively the emergency by predicting and alarming an epileptic fit. CONSTITUTION: A scalp brain wave detection portion measures scalp brain waves of an epileptic and outputs the measured data to a noise detection portion(S400). The noise detection portion selects a context window to set up a reference for deciding noise by using correlation density of electro-coulograph element and correlation density of electro-myograph(S402). The noise detection portion detects the noise of a corresponding window on the basis of the context window(S404). The next process is performed if a noiseless section passes a predetermined time, namely 10 seconds(S406,S408). A brain wave index calculation portion sets up a brain wave index by calculating a non-linear period of a phase pattern(S410). A fit symptom predication portion performs a calculating function and a deciding function for detecting a fit symptom(S412). A fit alarm portion performs an alarm function by using a predicated result(S414).