Концепція розробки інтелектуальної системи діагностики захворювань

Abstract

The development of the information technologies resulted in a creation of plenty of the medical informative systems from the automated ones to the intellectual systems of decision support. Co-operative complications at the decision making, absence of the terminology standardization and scales of measuring, badly formalized information and knowledge hamper the usage of the intellectual systems in the medical field. These resulted in a way that such systems are the least studied area of the artificial intelligence. The disease diagnostics intellectual system (DDIS) allows modeling of the doctor’s thinking process as it applies to the decision making in a specific task, to form certain conclusions, based on the knowledge which the system disposes, to explain achieved result. The development concept of DDIS, which allows the selection of the separate planning stages of the system, is offered in this article. On the stage «Task setting and preliminary research» is determined the purpose and the setting of the system, the expedience of its development and structure. The stage «Mathematical description of the system» is oriented to the database filling (DB) of the DDIS with private algorithms which produce the activity basis of the medic experts and cognitive engineers at the exposure of knowledge. At the stage «Construction of the disease diagnostics verbal model» a medical expert determines the possible amount of linguistic variables and their terms, the transmitters of fuzzy plural limitations for the values of linguistic variables, types and parameters of the membership functions. A cognitive engineer determines the set of fuzzy rules of the «IF... THEN»-type. At the stage «Abstract synthesis of the knowledge base» is carried out the check of knowledge base (KB) for the plenitude, noncontradiction and connectedness. The stage «Construction of the knowledge base» is a process of the rules formalization, described on the stage of the verbal model creation. The stage «Tuning of the knowledge base» is related to tuning of the fuzzy system: construction of the membership functions and forming of rules. On the stage «Minimization of the knowledge base» the number of rules is diminished with the preservation of the KB’s exactness. Research is conducted on the plenitude, noncontradiction and connectedness of KB. The stage «Estimation of the DDIS’s conceptual model» is related to the estimation of quality indicators of the system’s creation and functionality. And the last stage «Realization of the fuzzy model the disease diagnostics» supposes the program realization, testing and analysis of DDIS’s performance.