Predictive Analysis of Cardiovascular Disease under Double-Layer Attention Mechanism

Cardiovascular disease (CVD) is a major contributor to the current rise in mortality rates, so accurate prediction of CVD is crucial. Traditional prediction methods often rely on a single type of data, ignoring the complementarity between different modalities of information. This study proposed a multi-modal deep learning framework using ResNet-BiLSTM to extract ECG features and Graph Neural Networks (GNN) to extract clinical data features. Two static and dynamic attention mechanisms were added to optimize the fusion of multi-modal features. The experimental results show that the proposed method performs better than traditional models in terms of accuracy, F1 score, and AUC [1].