Diagnosis Prediction on MIMIC-IV
0.425MAR@20Clinical Mosaic (Projection)
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Clinical Mosaic (Projection)Embedding Generation=Projection layer2025.02 | 0.425 | 0.556 | 0.439 | 0.556 | 0.439 | 0.556 | — | |
| Clinical Mosaic (Concat)Embedding Generation=Concatenation2025.02 | 0.42 | 0.569 | 0.425 | 0.571 | 0.425 | 0.571 | — | |
| Clinical Mosaic (Mean)Embedding Generation=Mean pooling2025.02 | 0.416 | 0.538 | 0.423 | 0.567 | 0.423 | 0.567 | — | |
| Clinical GAN2025.02 | 0.41 | 0.558 | 0.414 | 0.559 | 0.414 | 0.559 | — | |
| Transformer Only2025.02 | 0.398 | 0.565 | 0.405 | 0.566 | 0.405 | 0.566 | — | |
| LIG-Doctor2025.02 | 0.267 | 0.474 | 0.361 | 0.431 | 0.42 | 0.402 | — | |
| Doctor AI2025.02 | 0.233 | 0.206 | 0.233 | 0.207 | 0.233 | 0.207 | — | |
| EHR2PathFine-Tuning protocol=Specialized Pathway Fine-Tuning (FT-P)2025.06 | — | — | — | — | — | — | 59 | |
| EHR2PathFine-Tuning protocol=Outcome-Oriented Fine-Tuning (FT-O)2025.06 | — | — | — | — | — | — | 59 |