Counterfactual Effectiveness on MIMIC Chest X-ray 192x192 (test)
0.03|ΔAUC| (Sex)Ribeiro et al.
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Ribeiro et al.Fine-tuning Protocol=w/o CF fine-tuning, Intervention=do(D = d)2025.10 | 0.03 | 20.14 | 0.56 | 22.01 | |
| Xia et al.Fine-tuning Protocol=Soft CF fine-tuning, Intervention=do(D = d)2025.10 | 0.03 | 1.54 | — | 3.69 | |
| OT-FLOWFine-tuning Protocol=None (zero-shot), Intervention=do(D = d)2025.10 | 0.067 | 0.627 | 0.435 | 2.28 | |
| Xia et al.Fine-tuning Protocol=Soft CF fine-tuning, Intervention=do(S = s)2025.10 | 0.07 | 2.04 | — | 0.09 | |
| Ribeiro et al.Fine-tuning Protocol=w/o CF fine-tuning, Intervention=do(R = r)2025.10 | 0.13 | 36.94 | 0.229 | 7.81 | |
| Ribeiro et al.Fine-tuning Protocol=w/o CF fine-tuning, Intervention=do(A = a)2025.10 | 0.13 | 20.44 | 3.872 | 6.11 | |
| Xia et al.Fine-tuning Protocol=Soft CF fine-tuning, Intervention=do(R = r)2025.10 | 0.13 | 3.36 | — | 0.11 | |
| OT-FLOWFine-tuning Protocol=None (zero-shot), Intervention=do(rand)2025.10 | 0.15 | 0.73 | 0.51 | 0.64 | |
| OT-FLOWFine-tuning Protocol=None (zero-shot), Intervention=do(S = s)2025.10 | 0.173 | 0.583 | 0.333 | 0.023 | |
| OT-FLOWFine-tuning Protocol=None (zero-shot), Intervention=do(R = r)2025.10 | 0.18 | 0.05 | 0.394 | 0.31 | |
| OT-FLOWFine-tuning Protocol=None (zero-shot), Intervention=do(A = a)2025.10 | 0.187 | 1.197 | 0.836 | 0.347 | |
| Ribeiro et al.Fine-tuning Protocol=w/o CF fine-tuning, Intervention=do(rand)2025.10 | 1.33 | 24.14 | 1.166 | 7.01 | |
| Ribeiro et al.Fine-tuning Protocol=w/o CF fine-tuning, Intervention=do(S = s)2025.10 | 7.43 | 20.84 | 0.486 | 1.31 |