Intelligence Diagnostics of Heart Disease Based on Neural Networks Ensemble Use

Abstract

The problem of constructing an intelligent system for diagnosing heart valve disease is considered. It is shown that the diagnosis is established on the basis of the results of a standard examination, which includes anamnesis, laboratory data, electrocardiogram, echocardiography and Doppler examination. The use of hybrid neural networks of ensemble topology is substantiated for solving the problem. In order to show that the appropriate neural networks for composing an ensemble can be effectively selected from a set of available neural networks, an algorithm for structural-parametric synthesis of hybrid neural networks of ensemble topology is proposed. As for training component neural networks it is used Bagging approach. Accuracy and diversity were used as criteria. The structure of the diagnostic system for recognition of valvular heart disease is presented. The results of research of the developed software are presented. As for training component neural networks, the Baging approach is used. The structure of the diagnostic system for recognition of valvular heart disease is presented. The results of the study of the developed software are presented.

TSK Fuzzy Neural Network Use for COVID-19 Classification

“…Ensembles are a powerful approach in machine learning, when combining several different models can improve the accuracy of the overall result. This approach is used to build models for solving problems in many areas [8], [9]. Combining several neural networks into an ensemble allows to get a model with less variance.…”