SOFTWARE MODULE FOR PRELIMINARY DIAGNOSIS OF PATIENTS BASED ON KOHONEN NEURAL NETWORK

Abstract

As you know, diagnosis is an extremely important aspect in the process of restoring health. When a patient seeks medical help with certain complaints, in most cases, each doctor will prescribe a general or extended (biochemical) blood test. This is a basic diagnostic procedure. A general analysis will allow establishing the corresponding fact of a violation in the body's work. Biochemical analysis of blood will provide more accurate information about the state of most vital organs, and will allow to evaluate the main metabolic processes. The results of the analysis are of high importance precisely at the stage of diagnosis, and, subsequently, when monitoring the recovery process. Monitoring is necessary if it is necessary to control the effectiveness of therapy. At the same time, the article deals with current and important issues of developing a software module for preliminary diagnosis of patients by blood analysis. Therefore, the time of execution and the speed of obtaining blood test results are important. The program module offered to your attention is based on the Kohonen neural network. Since such a neural network is a learning network, it becomes an excellent assistant in our task as a whole and in further diagnostics. Diagnosis is based on the results of the analysis, while a large number of important parameters are preserved with sufficiently fast operation of the algorithm. Therefore, this software module is necessary to increase the reliability of preliminary diagnosis, relative to competitors, and save time for doctors and patients. The article describes the structure, mathematical model and order of functioning of Kohonen's neural network. The architecture of the neural network software module is considered. An algorithm for the functioning of the software module and the ready-made application has been developed.