An Attention-Based Multi-Modal Network Framework for Cardiovascular Disease Detection and Grading
A novel Hierarchical Attention-based Representation learning with Multi-modal Network (HARM-Net) framework to grade the severity of CVD and the significance of each component and the specific contribution of cross-attention fusion to improved model performance for accurate, multimodal diagnosis of CVD is demonstrated.