N. Melnykova, U. MarikutsaFull text PDF

Abstract. In this article suggest some new approaches to solving the problems of medical data analysis and their personalization. To accomplish this was proposed to create a decision support system to the execution of sequence stages of analysis of patient’s data. The main stages of development and design of decision support systems that enable to make decomposition of control process and describe the relationship between input and output control flows. Applying the theory of decision trees during construction of decision trees of decision support system is due to the formation of a sequence of questions asked by the doctor when searching an individual approach when choosing a treatment. Decision tree creates a hierarchical structure of rules. This approach allows you to present the logic of sequence issues by doctor in solving the medical problem history and it makes possible to simulate decision making process by physician when selecting treatment scheme. Search the target value of output of decision medical support system makes it possible to select top of graph system that is located with more probability on the best way to the target. Important step in addressing the process of personalizing treatment schemes is estimated function that is based on Bayes theorem. Weight of occurrence next event corresponds to the highest value of the posterior probability of occurrence of the next state, given the time-dependent input parameters. Proposed improved method of decision-making for personalization standard schemes by modifying the method of decision-making based on decision trees considering relationship between the input parameters and evaluation function and result of its works is a personalized therapeutic scheme of treatment. It analyzed the quantitative results of applying the proposed method and existing for determining personalized schemes.

Key words: decision support system, data formalization, therapeutic treatment schemes, personalized of data.

Specifics personalized approach in the analysis of medical information

an international quarterly journal on economics of technology and modelling processes