Decision support system for anemia diagnosis based on fuzzy model

Abstract

Anemia is considered one of the most common diseases that affect the human body. In case of prolonged existence, it can lead to the development of chronic diseases due to excessive load on the vital organs. Effective treatments are available with early detection of the disease and largely depend on its underlying cause. In terms of severity, anemia can be divided into mild, moderate, and severe according to the level of hemoglobin. The morphological classification provides for the division of anemia into microcytic, normocytic, and macrocytic depending on the average volume of red blood cells. According to the hemoglobin, content anemia can be hypochromic, normochromic and hyperchromic. The article explores scientific and technical information on classification and fuzzy logic as well as describes the development of a decision support system for anemia diagnosis using a fuzzy logic inference model. During the analysis of existing literature, it was found that the use of naive Bayesian methods is optimal for solving the problem of determining anemia. But there is a possibility of minor errors in measuring features, so it is necessary to use the theory of fuzzy sets. Modulation of fuzzy logic inference was implemented using the Mamdani algorithm. The developed information system has two levels. The first level determines the severity of anemia. The second level is responsible for belonging to one of the classes: microcytic hypochromic, normocytic normochromic, and macrocytic. The software was written in Java on the basis of its own library to implement a computational algorithm for morphological classification of anemia and classification by severity according to the indicators of hematological laboratory tests. The application of the proposed architectural design of the system saves time for decision-making and eliminates the need for additional manpower to solve the problems of information and analytical support for management decision-making in support of business processes of anemia monitoring