A.V. Novikov, S.P. Seregin, L.V. Starodubceva

Different forms of pyelonephritis in structure of urology diseases. Their opportune prophylaxis is an important problem, it-s should be based on adequate selection of patients. That is why, invention of automated system of support of making decisions in prediction of progression pyelonephritis is actual. There was suggested informative signs of pyelonephritis, which presence in patients may promote progression of disease. Working vocabulary includes 55 informative signs. There are next group of signs: social, economical, productive, behavior, nutrional, medicobiologic. All signs describe influence on different factors of progression inflammation in upper urinary tract in concrete person. Using each informative signs as carrier of membership function, group of experts-specialists in urology under leadship engineer made modeling of membership function to the class ω1 (presence of risk of progression of pyelonephritis) ? μ ω1(xi+1). After it membership functions aggregate in final prognostic rule with calculation of coefficient of confidence in progression of pyelonephritis (КУОРП). To determinate liminal value КУОРП, which allowes to attribute observing patient in class ω1 was made retrospective analysis of known cases of year observation. On objects of control choices with well-known results were selected two classes: ω0 - no pyelonephritis, ω1- patient had pyelonephritis during the year. During dephazification of unclear conclusion were determined luminal values confidence coefficients equal 0,35, 0,55 and 0,75. Risk of progression of pyelonephritis dependes on coefficient of confidence may absent (<0,35), may be medium (0,35<КУ<0,55) or high (0,55<КУ<0,75). When value 0,75 and high should be decided question of diagnostics of pyelonephritis. On base of received information makes decision about conduction prophylaxis measures differ which dependes on risk level of disease progression. Verification efficacy of prophylaxis measures showed confidence to faultless work of received decision rules no less 0,95. Leading of patient under control of prediction system with using suggested prophylaxis methods, prevents progression of disease in patients with high risk in 80% of cases, and reduces quantity of recidivations of diseases to 8-15%.