Classification of cardiac arrhythmia using a convolutional neural network and bi-directional long short-term memory

Area of Science:

• Cardiology
• Artificial Intelligence
• Signal Processing

Background:

• Cardiac arrhythmia poses a significant global health risk, contributing to cardiovascular disease and high fatality rates.
• Early and accurate diagnosis of arrhythmias, characterized by irregular heart rates, is crucial for preventing strokes.
• Manual interpretation of electrocardiogram (ECG) signals for arrhythmia detection is time-consuming and prone to errors.

Purpose of the Study:

• To propose a novel deep learning model for the automatic classification of cardiac arrhythmias.
• To enhance the accuracy and efficiency of arrhythmia diagnosis compared to manual methods.
• To evaluate the model's performance across different arrhythmia classes and datasets.

Main Methods:

• Development of a hybrid deep learning model integrating Convolutional Neural Network (CNN) and Bi-directional Long Short-Term Memory (BiLSTM).
• Classification of five beat types: non-ectopic (N), supraventricular ectopic (S), ventricular ectopic (V), fusion (F), and unknown (Q).
• Training, validation, and testing of the model using the MIT-BIH and St. Petersburg arrhythmia databases.

Main Results:

• The proposed model achieved high performance on the MIT-BIH dataset, with training, validation, and testing accuracies of 100%, 98%, and 98%, respectively.
• The model demonstrated strong performance in classifying various arrhythmia types, measured by precision, accuracy, recall, specificity, and F1-score.
• Performance on the St. Petersburg dataset was lower than on the MIT-BIH dataset, indicating potential dataset-specific variations.

Conclusions:

• The novel deep learning model shows significant promise for accurate and automated cardiac arrhythmia classification.
• This approach offers a more efficient and reliable alternative to manual ECG interpretation for arrhythmia diagnosis.
• Further validation on diverse datasets is recommended to confirm generalizability and clinical applicability.