Proposal of analysis method focusing on nonlinear dynamics of human EEG with chaos

Abstract

Electroencephalogram (EEG) which has a chaotic fluctuation is difficult to analyze. However, quantitative analysis is sufficiently possible since EEG behavior is deterministic dynamics. Our method identifies EEG model parameters experimentally in consideration of chaotic dynamics of EEG. The purpose of this study is to examine the specific characteristics of model parameters. Validation of the method and investigation of characteristics of model parameters were conducted based on alpha frequency EEG data in the relax state and stress state. The results of the parameter identification with the time sliding window for 1 second, the nonlinear mathematical model is shown to produce outputs that can closely match the complicated experimental EEG data. Further, the results showed that the existence of nonlinear term in the EEG analysis is important and the linearity parameter shows a certain tendency as the nonlinearity increases. Furthermore, the activities of EEG become linear on the mathematical model when suddenly changing from the relax state to the stress state. Therefore, it is the effective analysis method that can calculate the degree of concentrate from the dynamics of EEG signal directly. The results suggest that our method may provide useful information in various field including the quantification of human mental or psychological state, diagnosis of brain disease such as epilepsy and design of brain machine interface.