Record-grounded Diagnosis on EMR Diagnosis 300 held-out cases (test)
4.3583ScoreLung-R1-14B
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Lung-R1-14BSize=14B2026.06 | 4.3583 | 82.75 | 71.3 | 89.39 | |
| Lung-R1-7BSize=7B2026.06 | 4.355 | 82.95 | 75 | 89.39 | |
| Claude-Sonnet-4.5Version=4.52026.06 | 4.2107 | 58.87 | 61.9 | 87.53 | |
| GPT-5.1Version=5.12026.06 | 4.2075 | 61.2 | 60.5 | 78.68 | |
| GPT-5-chatVariant=chat2026.06 | 4.1986 | 59.9 | 56 | 85.93 | |
| Gemini-3-ProVariant=Pro2026.06 | 4.0688 | 54.31 | 50.7 | 70.56 | |
| GPT-4.1Version=4.12026.06 | 4.0684 | 71.78 | 57.3 | 86.06 | |
| DeepSeek-V3.2Version=3.22026.06 | 4.0573 | 63.71 | 52.4 | 84.89 | |
| GPT-4oVariant=o2026.06 | 3.9908 | 67.58 | 53.5 | 84.93 | |
| DeepSeek-R1Variant=R12026.06 | 3.9891 | 71.29 | 48.9 | 81.03 | |
| Baichuan-M2Version=M22026.06 | 3.9339 | 72.71 | 50.8 | 81.96 | |
| GPT-OSS-120BSize=120B2026.06 | 3.9218 | 57.35 | 40.8 | 87.95 | |
| Qwen2.5-14BSize=14B2026.06 | 3.7709 | 73.22 | 41.5 | 79.31 | |
| ClinicalGPT-R1Variant=R12026.06 | 3.7449 | 74.52 | 47.6 | 79.58 | |
| MedGemma-27BSize=27B2026.06 | 3.7347 | 65.56 | 37.2 | 83.43 | |
| Qwen2.5-7BSize=7B2026.06 | 3.7007 | 81.16 | 47.1 | 72.81 | |
| HuatuoGPT-o1-7BSize=7B2026.06 | 3.3687 | 82.78 | 37.4 | 65.78 | |
| Qwen3-30B-A3BSize=30B2026.06 | 3.3617 | 69.04 | 44.7 | 82.49 | |
| Kimi-K2-ThinkingVariant=Thinking2026.06 | 3.3112 | 82.66 | 31.3 | 68.44 | |
| GPT-OSS-20BSize=20B2026.06 | 3.2727 | 70.26 | 24.2 | 79.05 |