Scholarlink

Abstract <p>This paper studies and applies methods for diagnosing cardiovascular diseases on the basis of electrocardiogram data using neural networks. We will use recurrent and convolutional neural networks. The key goal of this work is to build a neural network model for solving problems of classifying a set of various cardiovascular diseases. The initial data with which we will work and on the basis of which we will draw conclusions are one-dimensional electrocardiogram signals. These are waves with a duration of 10 s and a sampling frequency of 100 and 500 Hz. Also, on the basis of our signal, segmentation for this signal will be calculated. It&#xa0;displays the belonging of the signal element to one of the main peaks (segments) of the signal. As an extension of the sample and augmentation, we will not supply the entire signal, but its parts lasting 6 and 9 s, and will also periodically add normal noise. To evaluate the results of the model, we will use the following quality metrics: precision, recall, specificity, and F1-measure. The tasks we are considering are binary and multiclass classifications. We will see that many diseases described in the task are determined with fairly good metric values. We will also compare the results of each model. The work is designed to speed up and improve the work of doctors and help them in solving the problems of diagnosing heart pathologies.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Diagnostics of Cardiovascular Diseases Using Recurrent and Convolutional Neural Networks

  • 期刊论文

  • 发表日期：

    2025年6月25日

查看全文
  • V. V. Kuznetsov,
  • V. A. Moskalenko,
  • N. Yu. Zolotykh

摘要

Abstract

This paper studies and applies methods for diagnosing cardiovascular diseases on the basis of electrocardiogram data using neural networks. We will use recurrent and convolutional neural networks. The key goal of this work is to build a neural network model for solving problems of classifying a set of various cardiovascular diseases. The initial data with which we will work and on the basis of which we will draw conclusions are one-dimensional electrocardiogram signals. These are waves with a duration of 10 s and a sampling frequency of 100 and 500 Hz. Also, on the basis of our signal, segmentation for this signal will be calculated. It displays the belonging of the signal element to one of the main peaks (segments) of the signal. As an extension of the sample and augmentation, we will not supply the entire signal, but its parts lasting 6 and 9 s, and will also periodically add normal noise. To evaluate the results of the model, we will use the following quality metrics: precision, recall, specificity, and F1-measure. The tasks we are considering are binary and multiclass classifications. We will see that many diseases described in the task are determined with fairly good metric values. We will also compare the results of each model. The work is designed to speed up and improve the work of doctors and help them in solving the problems of diagnosing heart pathologies.