.png){kind=icon}
.png){kind=icon}
Return
MAGMED: A clinically grounded LLM-guided synthetic augmentation framework for admission-time ICU mortality prediction
.png){kind=icon}
PRE
AI
W
C
Z
X
J
DOI：10.1016/j.compbiomed.2026.111857.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• MAGMED integrates lightweight LLM guidance with VAE-based biomarker synthesis. • Among all evaluated methods, MAGMED demonstrated superior overall performance, with the highest external AUROC improved F1-score and recall. • Diffusion and conventional baselines were compared across internal and external cohorts. • Deployment-ready interfaces support clinician review and feedback loops.
Journal
IF:
6.3
Papers:
8.3K
Citations:
3.3W
Organization
Cited Papers