O.M. Grigorieva

Now technologies of the intellectual analysis of the data are widely applied at construction of systems of support of decision-making (SSDM) including in the field of medicine. Usually into structure SSDM enter: the block of preparation of the data (ETL-procedure), storehouse of the data, the block of the analysis and procedure of visualisation of the analysis results. Two kinds of the analysis are extended: OLAP and Data Mining. Data Mining allows to form two types of models: descriptive (can be presented clusters, associative rules) and predictive (to this type of models carry trees and tables of decisions, classification rules and regressional functions). Developers of the program systems realising technologies OLAP, originally focused the products on the decision of problems in spheres of business and the finance, and only rather recently there were the offers applying OLAP for the decision of problems of medicine, and in overwhelming majority of cases it is software of regional and federal level. The majority of offered systems use ROLAP realisation of storehouse of the data. Almost all software products realise only the OLAP-analysis, i.e. give analytics the information on which basis that can accept or deny own hypotheses. Procedures Data Mining are realised less often. The problem of diagnostics of diseases can be solved by construction of predictive models Data Mining in which basis qualifiers of objects lie. As sources of the information for construction of models act: medical devices, the patient and the doctor. Voice-frequency threshold audiogram is an example the static medical and biologic images presented by homogeneous discrete signs, set on set of the whole real numbers lying in a range from 0 to 110. By the form audiograms can be judged localisation of a pathology of an organ of hearing of the patient. As the initial data for the analysis audiograms of archive of Regional clinical hospital of Tver have served. The made computing experiments have shown that primary signs do not allow to receive successful splitting of objects into sample of the real clinical data. Rules of classification and system of binary signs which allow to allocate the classes corresponding to following conditions of an organ of hearing on analyzed sample are offered: Norm and three classes of a pathology. At calculation of signs it is used the adjusted elements which values are found empirically.