Table 1.
Number of records classed for five superclass disease categories in the PTB-XL dataset.
Fig 1.
Structural flowchart of the multimodal cross-attention fusion network MAF-Net model.
Fig 2.
Model structure flowchart of the X branch (clinical data processing branch) in the MAF-Net model.
Fig 3.
Model structure flowchart of the Y branch (ECG data processing branch) in the MAF-Net model.
Fig 4.
Feature fusion model structure flowchart for the X branch (clinical data) and Y branch (ECG data) in the MAF-Net model.
Table 2.
Performance metrics comparison between our MAF-Net model and various methods (mean ± standard deviation).
Table 3.
Ablation experiment results for the MAF-Net model (mean ± standard deviation).
Fig 5.
Model recognition accuracy of MAF-Net after removing key clinical features on the PTB-XL dataset.
Table 4.
Comparison results of key clinical feature ablation (mean ± standard deviation).
Fig 6.
Comparison of ROC curves for the MAF-Net model (with and without X branch) across five hyperclasses on the PTB-XL dataset.
Fig 7.
Confusion matrices for the MAF-Net model (with and without the X branch) across five hyperclasses on the PTB-XL dataset.
Fig 8.
Bar chart comparing different metrics across five hyperclasses for the MAF-Net model (without X branch) on the PTB-XL dataset.
Fig 9.
Bar chart comparing different metrics across five hyperclasses for the MAF-Net model (with X-branch) on the PTB-XL dataset.
Fig 10.
t-SNE visualization of spatial clustering for five hypercategories on the PTB-XL dataset using the MAF-Net model (with and without the X branch).