NEURAL NETWORK MODELING OF THE PROCESS OF SELECTING A PATTERN FOR THE TREATMENT OF PATIENTS WITH CHRONIC PYELONEPHRITIS AND UROLITHIASIS

The article deals with the basic aspects of designing a neural network model for choosing a treatment regimen for chronic pyelonephritis and urolithiasis. One of the most common non-specific chronic kidney diseases is a chronic pyelonephritis. Currently, mathematical modeling of biological systems is one of the main directions of mathematical methods in medical practice. The paper demonstrates network operation. The construction of a multilayer perceptron was carried out on the basis of the Neural Networks module in the Statistica program. The resulting neural network model has 5 outputs, each of which is identical to the types of treatment present in the training set. The developed model provides an opportunity to choose one of 5 types of treatment: Y1 - conservative therapy with antibacterial, antispasmodic and anti-inflammatory drugs in combination with physiotherapeutic procedures; Y2 - conservative therapy in combination with surgical treatment in the amount of contact lithotripsy (KLT); Y3 - conservative therapy in combination with surgical treatment in the volume of distant lithotripsy (DLT); Y4 - conservative therapy in combination with surgical treatment in the amount of percutaneous nephrolitholapaxy (PNLT); Y5 is an open surgery and conservative treatment. The developed model makes it possible to choose one of 5 types of treatment. The reliability of this model was 94%.

1. Savas L. Nosocomial urinary tract infections: micro-organisms, antibiotic sensitivities and risk factors. / L. Savas, S. Guvel, Y. Onlen//WestIndianMed. -2006.-Т. №55 (3). -р. 188-193.

2. Urology: National Leadership / Ed. ON. Lopatkina. - M.: GEOTARMedia, 2009.- P.434-451.4.

3. Korovin E.N., Rodionov O.V. Methods of processing biomedical data: a training manual. Voronezh: VSTU, 2007.

4. Novikova E.I. Algorithmization and management of the process of diagnostics of gynecological diseases based on multivariate modeling / E.I. Novikova, O.V. Rodionov // monograph. Voronezh: VSTU, 2012. 132с.

5. Novikova E.I. Simulation of biomedical systems / E.I. Novikova, O.V. Rodionov, E.N. Korovin // study guide, Voronezh: VSTU, 2008. - 196s.

6. Korovin E.N., Levenkov K.O., Ryabchunova L.V. Analysis and algorithmization of diagnostic processes and the choice of tactics for the treatment of chronic pyelonephritis based on simulation modeling // System analysis and management in biomedical systems. - 2016. V. 15. № 1. S. 84-87.

7. Levenkov K.O. Cluster analysis of data on the choice of tactics for the treatment of chronic pyelonephritis / K. O. Levenkov, A.S. Turbin, E.N. Korovin, A.V. Kuzmenko, T.A. Gyaurgiev // System analysis and management in biomedical systems. -2017.- T. 16. No. 4. -p.857-861.

8. Korovin V.N. Developing a neural network for diagnosing chronic pyelonephritis / V.N. Korovin, E.N. Korovin, K.O. Levenkov, M.V. Lushchik // System Analysis and Control in Biomedical Systems. -2015. - T. 14. No. 3. - p. 585-588,647-651

9. Korovin E.N. Intellectualization of the process of diagnosing chronic pyelonephritis based on a priori ranking of expert opinion / E.N. Korovin, V.N. Korovin, K.O. Levenkov, M.V. Lushchik // System Analysis and Control in Biomedical Systems. - 2016. -T. 15. No. 4.-p. 647-651.

10. Levenkov K.O. Digital processing of the results of Doppler ultrasound in patients with chronic pyelonephritis and urolithiasis / K.О. Levenkov, E.N. Korovin // System analysis and management in biomedical systems. - 2018. -T.17. Number 3. - p. 686-692.