Binary disease diagnosis on VinDr-CXR OOD
93.5Macro AccuracyCSRL
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CSRL2026.05 | 93.5 | 48.2 | |
| Majority Voting (tie → neg)2026.05 | 93.4 | 40.9 | |
| Gemini-3.1-Pro-Preview2026.05 | 93 | 45.5 | |
| DenseNet-CheX2026.05 | 91.6 | 27.1 | |
| CheXzero2026.05 | 90.4 | 36.2 | |
| MedGemma-27B-IT2026.05 | 90.2 | 40.3 | |
| MedKLIP2026.05 | 89.7 | 38.3 | |
| HuatuoGPT-Vision-7B2026.05 | 89.4 | 28.2 | |
| Majority Voting (tie → pos)2026.05 | 87.5 | 33.3 | |
| Hulu-Med-Flash-Preview-27B2026.05 | 86.8 | 41.1 | |
| GPT-4o2026.05 | 85.1 | 20.1 | |
| Best Tool per Pathology2026.05 | 82.2 | 37.4 | |
| Logistic Regression2026.05 | 82.1 | 36.9 | |
| Qwen2.5-VL-7B-Instruct2026.05 | 81.7 | 12.1 | |
| DenseNet-MIMIC2026.05 | 78.7 | 17.9 | |
| Lingshu-7B2026.05 | 77 | 24.8 | |
| InternVL3.5-38B-Instruct2026.05 | 74.6 | 34 | |
| DenseNet-NIH2026.05 | 73.2 | 15.7 | |
| Claude-Sonnet-4.62026.05 | 54.3 | 12.9 | |
| CheXagent-8B2026.05 | 50.2 | 22.5 | |
| Qwen2.5-VL-72B-Instruct2026.05 | 43.5 | 10 |