A.A. Desova, J.A. Dorofeyuk, V.V. Gychyk

The automated methods of pulse signal analysis should provide the algorithms elaboration, realizing signal form analysis as well as signal rhythmical structure. The approach to signal informative characteristic detection, based on the sequential implementation of different algorithms in accordance with specific signal features, is considered. In particular, the concrete form of quasiperiod determines the formed local indices quantity and accordingly the dynamic series quantity, determining the signal rhythmical structure in whole. In this connection at the first stage the quasiperiod form is estimated and subsequent to the results of estimation the signal rhythmical structure formation algorithm is implemented. The investigations were carried out on the basis of the statistical material, received during clinical researches of children and teenagers, which were focused on the initial stage arterial hypertension detection. More than 400 patients with a variety of functional disturbance, aged from 9 to 16 years, were examined. All types of diseases were divided into two classes: first class (149 patients) ? initial arterial hypertension, second class (253 patients) ? various types of psychosomatic functional pathology along with the normal arterial tension. In the examined data array (403 vectors) the typological analysis of the signal oscillation unit form was performed. The automatic classification of signal form type on the basis of signal parameter computation was implemented. The set of 8 features, ensuring high quality of the investigated signal form automatic classification, is formed. The diagnostic estimations for different signal form types are received. It is shown, that the results of signal form type estimation determinate the way of signal rhythmic structure formation. The typological analysis of spectral density form of the dynamic series formed on different signal unit oscillation indices is implemented. The spectral density type formalized classification in the space of parameters - mutual relations of the spectral density peaks - was carried out. The comparative analysis of spectral density types for different dynamic series are is introduced. The relationship between the spectral density form type of the quasiperiod duration dynamic series and the diagnostic significance of the dicrotic wave duration dynamic series spectral density parameters is shown. It is shown, that in the context of investigated problem - the initial stage arterial hypertension detection in infancy and adolescence period - the most informative are those indices, which are determined by pulse signal dicrotic wave oscillatory components.