AP14969403-Study of methods for predicting myocardial infarction using machine learning algorithms
Currently, the existing classical methods and diagnostic tools based on the amplitude-time analysis of ECS do not meet the modern requirements for accurate diagnosis of MI (3 out of 4 myocardial infarctions are diagnosed). Given the existing imperfect devices and systems for diagnosing MI, as well as the human factor, it is obvious that there is a need to create a decision support system with increased accuracy in diagnosing MI that can help the doctor. The problem of increasing the reliability of diagnostics of the heart condition makes it necessary to improve methods for obtaining new diagnostic information.
Most of the existing research recommends machine learning (ML) algorithms for predicting heart disease. Accordingly, this paper develops a method for diagnosing the state of the heart, which is carried out by analyzing the output signals of neural networks trained to recognize direct and reciprocal signs of MI in the ECS of 12 generally accepted leads, and constructing decision rules. To do this, they perform: dividing the surface of the left ventricle of the heart into areas for which the detected direct and reciprocal signs of myocardial infarction (MI) correspond to a specific stage and a specific type of MI by the depth of the lesion; determining the stage of MI and the type of MI by the depth of the lesion by heart areas; determining the localization of the established stages. EFFECT: invention provides formation of a diagnostic report on the patient's heart condition, in which, in the presence of myocardial infarction (MI) of the left ventricle of the heart, its localization, stages of development (acute, acute, subacute, cicatricial) and types by depth of damage to the heart wall (transmural, subepicardial, subendocardial) are indicated, as well as a complete and accurate assessment of the heart condition the patient regardless of the doctor's level and work experience.
Scientific supervisor: PhD, Alimbayeva Zhadyra Nurdauletovna
Quantitative and qualitative composition of project performers: 6 performers, consisting of: 1 PhD, 2 PhD, 2 PhD doctoral students, 1 Master
Terms of implementation: 2022-2024