Classification of Arrhythmia in Heartbeat Detection Using Deep Learning.

Area of Science:

• Cardiology
• Artificial Intelligence
• Biomedical Engineering

Background:

• Electrocardiogram (ECG) is a vital diagnostic tool in healthcare.
• Deep learning shows significant potential for analyzing ECG data and predicting health outcomes.
• Arrhythmia detection remains a critical challenge in cardiovascular medicine.

Purpose of the Study:

• To apply and evaluate deep learning techniques for classifying cardiac arrhythmia using public ECG datasets.
• To compare the performance of Convolutional Neural Network (CNN), CNN+Long Short-Term Memory (LSTM), and CNN+LSTM+Attention models in arrhythmia classification.

Main Methods:

• Utilized two public ECG datasets: MIT-BIH Arrhythmia Database and PTB Diagnostic ECG Database.
• Implemented and compared three deep learning architectures: CNN, CNN+LSTM, and CNN+LSTM+Attention.
• Trained models on 80% of the data and tested on the remaining 20%.

Main Results:

• The CNN model achieved an accuracy of 99.12%.
• The CNN+LSTM model demonstrated a high accuracy of 99.3%.
• The CNN+LSTM+Attention model reached an accuracy of 99.29%.

Conclusions:

• Deep learning models, particularly CNN+LSTM, show excellent performance in classifying arrhythmia from ECG data.
• These findings suggest the utility of advanced AI techniques for improving the accuracy and efficiency of cardiac arrhythmia diagnosis.