2026-10-14T06:09:48Z https://api.figshare.com/v2/oai

oai:figshare.com:article/34027599 2026-09-29T20:22:08Z category_24559 item_type_2 month_year_09_2026

<b>Siamese-HeteroGNN</b><b>_</b><b> </b><b>Multimodal Graph Neural Network</b> Meiru Wu (25138863) Clinical sciences not elsewhere classified Echocardiographic graph neural network <p dir="ltr">This work contributes a patient-level heterogeneous graph modeling cross-modal relationships without concatenation, Siamese branches with contrastive loss, adaptive modality dropout with Monte Carlo uncertainty estimation, a SHapley Additive exPlanations (SHAP)-guided one-vs-rest nomogram, and ComBat-harmonized features validated on an independent held-out cohort built entirely on public data.</p> 2026-09-29T20:22:08Z Dataset Media 10.6084/m9.figshare.34027599.v1 https://figshare.com/articles/media/_b_Siamese-HeteroGNN_b_b_b_b_b_b_Multimodal_Graph_Neural_Network_b_/34027599 CC BY 4.0