血管学：开放获取
开放获取

抽象的

Enhancing Early Detection of Cardiovascular Diseases through Deep Learning-Based ECG Signal Classification

Immaculate Joy Selvam*, Moorthi Madhavan, Senthil Kumar Kumaraswamy

The global impact of Cardiovascular Diseases (CVDs) is profound and requires urgent attention. Accurately classifying heartbeats is essential for assessing cardiac function and detecting any irregularities. Electrocardiograms (ECGs) play a critical role in diagnosing CVD by providing graphical representations of the heart's electrical activity. In this study, Deep Learning (DL) models are employed to automatically categorize ECG data into different classifications, including normal, conduction disturbance, ST/T change, myocardial infarction, and hypertrophy. To achieve this classification task, we utilize the PTB-XL database, which eliminates the need for real-time patient data collection by providing a comprehensive collection of ECG recordings. To aid in the accurate classification of ECG data, we suggest a hybrid DL method that makes use of Convolutional Neural Network (CNN) architecture combined with Variational Autoencoder (VAE). We also evaluate our approach in comparison to other transfer learning methods that have shown potential in DL applications, such as ResNet-50 and Inception-v3. The suggested CNN- VAE architecture not only showcases superior accuracy but also provides satisfactory computational performance, contributing to the timely and automated classification of ECG signals. This advancement plays a important role in the early detection and effective management of CVDs, thereby enhancing healthcare outcomes for individuals at risk.