Hybrid Deep Learning Models for Accurate ECG Classification in Cardiovascular Disease Diagnosis

Electrocardiograms (ECGs) are very important for diagnosing Cardiovascular Diseases (CVDs) because they record important information about the electrical activity of the heart. As CVDs are so common and fatal, it is very important to predict them promptly using ECG signals. This study suggests a new type of hybrid deep learning architecture called CRNN (Convolutional Recurrent Neural Network), which combines the ability of CNNs to extract spatial features with the ability of LSTMs to model temporal one. The MIT-BIH Arrhythmia Database was used to train and test the model in classifying ECG beats. The proposed model uses convolutional operations, LSTM state transitions, and a softmax classification function. The CRNN does better than CNN or LSTM models on their own, with high classification accuracy and low false positive rates. This study shows that hybrid deep models can be used for robust ECG analysis and clinical decision support systems.