N.A. Korenevsky, F.Ionescu, A.A.Kuzmin, R.T. Al-Kasasbeh

The analysis of the data structure used in applications of automated medical diagnostics and forecasting, allows to draw a conclusion on expediency of use the theory of decision-making fuzzy logic as the basic mathematical method. For the purpose of improvement of decision-making quality we suggest to unite two approaches which are accepted in the classical fuzzy logic. The first approach is on the basis of the logic conclusions which make with use of membership functions. The second approach is on the basis of definition of confidence factor through belief and disbelief measures to investigated classes of conditions. We suggest defining a kind, parameters of membership functions and ways of their aggregation with confidence factors on the basis of exploration analysis' methods. At such approach synthesis of fuzzy solving rules is carried out in three stages. At the first stage we make the exploration analysis which allows studying geometrical structure of classes in space of informative factors. We term interposition of various classes - objects on training sample as structure. At the second stage we choose universes of discourses and parameters of membership functions for known structure of classes and types of signs. These membership functions solve classification problems in subspaces and in areas of initial factor space. The choice is carried out so that each membership function provided the greatest possible confidence of classification or forecasting on each technological step of decision-making at the set complexity of the classifier. At the third stage the membership functions are united in private (on groups of the same factors and subspaces) and final solving rules which provide demanded confidence of accepted decisions. It allows considering structure of data at synthesis of fuzzy solving rules and it creates preconditions for optimization of decision-making processes. We were solved problems of forecasting, early and differential diagnostics at diseases of a gastroenteric path, cardiovascular system, respiratory system, thromboses of the central vein of a retina and its branches, atrophies of an optic nerve, urological diseases, anemias, the diseases caused by professional work and a number of others by means of the described mechanism of synthesis of fuzzy solving rules. The confidence is reached not worse 0,85 in forecasting problems. Also the confidence is reached not worse 0,9 in diagnostics problems. Such results allow recommending the offered approach of synthesis of fuzzy solving rules to use in medical and ecological systems of support and decision-making.