[An application of autoregressive model to pattern discrimination of brain electrical activity mapping].

We developed a method of illustrating significant regional differences between two records of brain electrical activity mapping (BEAM), using pattern discrimination of autoregressive (AR) EEG analysis. In this study, the EEG data could be objectively reduced to a few AR coefficients, and this statistical method was successfully applied to discrimination of differences in BEAM among records from two or more subjects. A significant difference was illustrated at the right parietal region between two age groups, 6 years and 8 years, in normal children. It was concluded that this new approach of BEAM analysis through a statistical taxonomy was clinically useful for evaluation of the normal development of BEAM in children and for localization of functional brain abnormalities.