Arrhythmia Classification Using Hybrid CNN-LSTM Model

Electrocardiogram (ECG) analysis is crucial for diagnosing and monitoring cardiovascular diseases. Accurate ECG signal classification is essential for early detection and intervention. This paper proposes a novel hybrid deep learning architecture for automated ECG signal classification using the widely known MIT-BIH Arrhythmia dataset. The architecture combines the strengths of Convolutional Neural Networks (CNNs) for extracting spatial features from ECG waveforms and Long Short-Term Memory (LSTM) networks which will be used in capturing the temporal dependencies within the sequential data. This combined approach aims to achieve superior performance in ECG classification compared to models relying solely on CNNs or LSTMs. The CNN-LSTM model is trained as well as evaluated on the MIT-BIH Arrhythmia dataset, achieving an accuracy of 98.94% surpassing several existing methods. This study contributes to the advancement of automated ECG analysis and offering potential improvements in cardiovascular disease diagnosis along with patient monitoring.