DIAGNOSTICS BASED ON BLOOD ANALYSIS INDICATORS USING THE ADABOOST ALGORITHM

Abstract

This article discusses the use of the AdaBoost algorithm for disease diagnosis based on blood parameters. The study analyzes how the algorithm adapts to various datasets consisting of blood parameters and identifies key biomarkers that affect the accuracy of diagnosis. It is shown that the use of this algorithm allows for increased pathology recognition efficiency compared to traditional methods, providing higher sensitivity and specificity. The abstract also includes a comparative analysis of AdaBoost performance with other machine learning models, highlighting its advantages in the field of diagnostics based on medical data. A systematic approach to the medical diagnosis process, methods, models and algorithms for making diagnostic solutions have been developed. The developed model and algorithms make it possible to create a system that uses the adoption of a hybrid intelligent diagnostic solution. A multivariate probabilistic model was created taking into account the weighting coefficient of experts and the mutual compatibility of experts' assessments. This allows to make a collegial diagnostic solution with a certain probability that the patient has the suspected disease. A generalized logical model of the multi-stage reasoning process of experts on diagnosis was created.