Additional opportunities for using computer technologies in expert analysis of laboratory data

Abstract

The presented work reflects the author’s algorithm of mathematical processing and possible interpretations of the results obtained when creating expert analytical systems using computer technology. As initial parameters for the construction of neural networks of the second level, it is proposed to use individual results of calculations of the ratios of a number of laboratory indicators capable of forming a single functional cluster (leukogram, indicators of water-electrolyte balance, protein fractions, etc.). Further, using the proposed calculation algorithm and the archive database, structural features in the ′deformation′ of the ratio panel were compared against the background of the growth of each indicator, a matrix table was built reflecting the degree (correlation coefficients, KCr) of the coincidence of the features of the formation of the structure of the obtained panels of the parameters determined in individual observations. At high values of KCr, a conclusion is made about their (determined parameters) unified participation in the mechanism of emerging disorders. Thus, the system allows you to establish a leading complex of associated relationships by changes in the structure of ratios in the selected panel, corresponding to the dynamics of deviation of the target (analyzed) indicator, thereby differentiating the different mechanisms of formation of pathological deviations in this patient, the manifestation and balance of their values in the system-wide response, to evaluate the specific value of the absolute indicator as the final result of such interaction at the time of examination of the patient. The approach used, proposed by the authors, allows us to evaluate not only quantitative relationships, but also the features of the manifestation of the functional properties of the evaluated indicators. All of the above significantly expands the informativeness of the obtained laboratory data, allowing us to build a reasonable paradigm of the connections of the emerging complex of pathological disorders in each individual case. In the list of references, the authors cite publications as concrete examples of the use of the proposed approach in assessing a complex of disorders with different pathological processes using different panels in calculations.