M.V. Artemenko - Ph. D. (Biol.), Associate Professor, Southwest State University, Kursk E-mail: artem1962@mail.ru E.S. Podvalny - Dr. Sc. (Eng.), Professor, Voronezh StateTechnical University, Voronezh E-mail: spodvalny@yandex.ru E.A. Startsev - Post-graduate Student, Department of Biomedical Engineering, Southwest State University, Kursk E-mail: starcev_evgeniy@mail.ru

The main purpose of the automated systems of support of decision-making in biomedical and ecological systems is the classification of the biological object and predict its behavior. As an adequate description of the behavior of the biological object as a response to changes in the internal and external environments external and autonomous control systems, reflected by the values of the recorded latent or observed and controlled variables - signs, the adequacy of forming of informative features is an urgent task. The study of biotechnical systems is characterized by the features of theirbehaviour as living, open, complex, and hierarchical structures are constantly exposed to the environment and with an autonomous control system having the properties of self-organization and self-regulation. This causes problems in determining an adequate informative value of features and their relevant ranking for building systems of support of decision-making in biology, medicine and ecology. Primarily, these include: difficulties in the organization ofreproducibility of experience, multimodality composition of the detected signs, \"rude\" and / or blurred values of the measured variables, a statistically small sample sizes, lack of structured data also, blurry and fuzzy crossing the boundaries of diagnostic classes, etc. In this regard, we consider the method of estimating the information content of the formation and composition of biotechnical characteristics of the studied system, based on the calculation in different ways values of the integral criterion of informativeness by applying to the analyzed data, tested the algorithms, based on: measurement theory, latent variables, structural-parametric self-organizing identification of nonlinear discriminant functions in the form of a polynomial of Kolmogorov-Gabor method of GMDH (group method of data handling), theories of qualitative expert analysis. Discusses the various methods of aggregation of partial indicators of informativeness in the calculation of the integral criterion of ad-ditive-weight, majority aggregation, \"careful\", \"tense\", \"greedy\" ? using the softmax operation, and analog-to-Boolean aggregation. In the process of synthesis of fuzzy models and inference rules for knowledge bases for automated decision-making systems the choice of form and parameters of the membership functions is poorly formalizable. This paper proposes the use of the considered values of partial indicators and integrated criterion of informativeness as estimates of the extremes of the membership functions after bringing them to the appropriate interval. Proposed new nonparametric methods of formation of informative signs biotechnical systems allow a semi-structured imprecise data and to synthesize diagnostic decision rules knowledge bases in various segments of the automation of decision makers on the basis of the achievements of artificial intelligence.

 

1. Voroncov I.M., SHapovalov V.V., SHerstjuk JU.M. Zdorove. Opyt razrabotki i obosnovanie primenenija avtomatizirovannykh sistem dlja monitoringa i skrinirujushhejj diagnostiki narushenijj zdorovja. SPb.: OOO «IPK «Kosta» B. 2006. 432 s.
2. Demidova L.A., Kirakovskijj V.V., Pylkin A.N. Prinjatie reshenijj v uslovijakh neopredelennosti. 2-e izd., pererab. M.: Gorjachaja linija - Telekom. 2015. 283 s.
3. Zagorujjko N.G., Kutnenko O.A. Algoritm GRAD dlja vybora informativnogo podprostranstva priznakov. Institut Matematiki SO RAN, (tochka dostupa http://pandia.ru/text/78/248/79351.php).
4. Kolesnikova S.N. Metody analiza informativnosti raznotipnykh priznakov // Vestnik Tomskogo gosudarstvennogo universiteta. Ser. Upravlenie, vychislitelnaja tekhnika i informatika. 2009. № 1(6). S. 48-54.
5. Korenevskijj N.A., SHutkin A.N., Bojjcova E.A. Ocenka i upravlenie sostojaniem zdorovja na osnove modelejj G. Rasha // Medicinskaja tekhnika. 2015. № 6.  S. 37-40.
6. Rasch G. Probabilistic models for some intelligence anent tests (Expanded edition, with foreword and afterword bu Benjamin D. Wright). Chicago: University of Chicago Press. 1980. 199 p.
7. Paket RUMM2020/ URL:http://rumm2020.software.informer.com/
8. Orlov A.A. Principy postroenija arkhitektury programmnojj platformy dlja realizacii algoritmov metoda gruppovogo ucheta argumentov // Upravljajushhie sistemy i mashiny. 2013. № 2. S. 65-71.
9. GMDH //URL: http://gmdh.net/ (data obrashhenija 8.02.2016).
10. Artemenko M.V., Kalugina N.M., SHutkin A.N. Formirovanie mnozhestva informativnykh pokazatelejj na osnovanii approksimirujushhego polinoma Kolmogorova-Gabora i maksimalnogo gradienta funkcionalnykh razlichijj // Izvestija JUgo-Zapadnogo gosudarstvennogo universiteta. Ser. Upravlenie, vychislitelnaja tekhnika, informatika. Medicinskoe priborostroenie. 2016. № 1. S. 116-123.
11. Satton R.S., Barto EH.G. Obuchenie s podkrepleniem. M.: BINOM. Laboratorija znanijj. 2011. 369 s.
12. Cifrovaja obrabotka signalov i izobrazhenijj v radiofizicheskikh prilozhenijakh / Pod red. V.F. Kravchenko. M.: Fizmatlit. 2007. 544 s.
13. Korenevskijj N.A., Artemenko M.V., Provotorov V.JA., Novikova L.A. Metod sinteza nechetkikh reshajushhikh pravil na osnove modelejj sistemnykh vzaimosvjazejj dlja reshenija zadach prognozirovanija i diagnostiki zabolevanijj // Sistemnyjj analiz i upravlenie v biomedicinskikh sistemakh. 2014. T. 13. № 4. S. 881-886.
14. Korenevskijj N.A., Razumova K.V. Sintez nechetkikh klassifikacionnykh pravil v mnogomernom prostranstve priznakov dlja medicinskikh prilozhenijj // Izvestija JUgo-Zapadnogo gosudarstvennogo universiteta. Ser. Upravlenie, vychislitelnaja tekhnika, informatika. Medicinskoe priborostroenie. 2012. № 2. S. 223-227.