The article presents a physical and mathematical model developed by the authors. The model allows analyzing the amplitude-frequency characteristic (AFC) of electroencephalograms (EEG) of a human brain. The proposed method of EEG processing, in contrast to the previously used one, is based on approximating the AFC by an equation that contains a set of coefficients. This method is convenient for comparing data obtained from different subjects and, as found, has diagnostic significance. Previously, only the frequency / amplitude ratio or its inverse was used for evaluations. The results achieved indicate the possibility of differentiating patients of various neuropsychic profiles according to the values of the parametric indicators obtained by approximating the EEG amplitude-frequency response.