A Comprehensive CNN-LSTMA Model for Cardiovascular Disease Diagnosis

Cardiovascular disease (CVD) continue to be one of the primary causes of mortality globally, necessitating accurate and early detection through advanced predictive models. This research introduces a novel hybrid Convolutional Neural Networks (CNN) - Long Short Term Memory (LSTM) model with Attention Mechanism, leveraging the strengths of CNN for spatial feature extraction, LSTM for temporal sequence processing, and an Attention mechanism for focusing on the most critical time steps and features. The model is evaluated on two widely recognized datasets, the UCI Cleveland and UCI Hungarian heart disease datasets. The experimental findings show notable enhancements in essential performance indicators, with the proposed model achieving an accuracy of 94.5% and Log-Loss of 0.143 on the UCI Cleveland dataset. The hybrid model not only outperforms traditional CNN, LSTM, and other deep learning models but also exhibits enhanced model robustness and generalization on imbalanced datasets. These improvements highlight the potential of this hybrid approach in providing more reliable and interpretable CVD predictions, offering valuable insights for clinical decision-making.