In-hospital mortality prediction on Clinical Multimodal Dataset (test)
0.89AUROCHAIM
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| HAIM2025.10 | 0.89 | 0.54 | |
| MedM2TAblation Configuration=Full Model2025.10 | 0.868 | 0.47 | |
| MedM2TAblation Configuration=w/o Pre-trained Encoder2025.10 | 0.868 | 0.477 | |
| MedM2TAblation Configuration=w/o Bi-Modal Attention2025.10 | 0.868 | 0.476 | |
| MedM2TAblation Configuration=w/o Shared Encoder2025.10 | 0.868 | 0.472 | |
| MultiModNEncoder Type=Our Encoder2025.10 | 0.867 | 0.455 | |
| MultiBenchFusion Strategy=LF, Encoder Type=Our Encoder2025.10 | 0.862 | 0.46 | |
| MultiModNEncoder Type=Original2025.10 | 0.856 | 0.409 | |
| MultiBenchFusion Strategy=LF, Encoder Type=Original2025.10 | 0.833 | 0.418 | |
| MultiBenchFusion Strategy=LRTF, Encoder Type=Original2025.10 | 0.823 | 0.368 | |
| MultiBenchFusion Strategy=LRTF, Encoder Type=Our Encoder2025.10 | 0.768 | 0.343 |