Clinical Structured Data Extraction on n2c2
101True Positives (TP)Infherno
Evaluation Results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| InfhernoLLM=Gemini-2.5 Pro2025.07 | 101 | 0 | 66 | 100 | 60.5 | 75.4 | 6 | 94.4 | 97.1 | 104 | 104 | 104 | |
| InfhernoLLM=DeepSeek V3.1 Chat2025.07 | 82 | 0 | 85 | 100 | 49.1 | 65.9 | 25 | 76.6 | 86.8 | 84 | 84 | 82 | |
| InfhernoLLM=Claude Sonnet 4.52025.07 | 78 | 0 | 91 | 100 | 46.2 | 63.2 | 31 | 71.6 | 83.4 | 79 | 78 | 78 | |
| InfhernoLLM=Qwen3-235B-A22B-25072025.07 | 76 | 0 | 92 | 100 | 45.2 | 62.3 | 32 | 70.4 | 82.6 | 77 | 76 | 76 | |
| InfhernoLLM=GPT-52025.07 | 68 | 0 | 100 | 100 | 40.5 | 57.6 | 40 | 63 | 77.3 | 69 | 63 | 42 | |
| InfhernoLLM=Qwen3-8B2025.07 | 32 | 0 | 134 | 100 | 19.3 | 32.3 | 74 | 30.2 | 46.4 | 36 | 36 | 35 |