Differential Diagnosis on DDxBench n = 150
80Top-1 Accuracy (Primary Diagnosis)PathChat+
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| PathChat+Input=Expert-curated ROIs2025.06 | 80 | 92 | |
| PathChat 1Input=Expert-curated ROIs2025.06 | 72 | 88.7 | |
| Gemini 2.5 proInput=Expert-curated ROIs2025.06 | 51.3 | 71.3 | |
| GPT-5-miniInput=Expert-curated ROIs2025.06 | 43.3 | 67.3 | |
| Claude Sonnet 4Input=Expert-curated ROIs2025.06 | 42.7 | 56 | |
| GPT-5Input=Expert-curated ROIs2025.06 | 40.7 | 61.3 | |
| Qwen3-VLInput=Expert-curated ROIs2025.06 | 38.7 | 62.7 | |
| HuatuoGPT-VisionInput=Expert-curated ROIs2025.06 | 33.3 | 54 | |
| LLaVA-OneVisionInput=Expert-curated ROIs2025.06 | 10.7 | 16.7 |