Appendicitis Diagnosis on Düsseldorf cohort
80.7AccuracyClaMPAPP
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| ClaMPAPPBackbone=Llama-3.1-8b, Classifier=XGBoost, Architecture=Hybrid LLM-as-interface, ML-as-predictor2026.06 | 80.7 | 93.5 | 39.4 | 83.3 | 88.1 | 215 | 28 | 43 | 15 | |
| OASST-Pythia-12b2026.06 | 73.1 | 92.2 | 11.3 | 77.1 | 84 | 212 | 8 | 63 | 18 | |
| OpenAI o3-pro2026.06 | 72.4 | 70.9 | 77.5 | 91.1 | 79.7 | 163 | 55 | 16 | 67 | |
| OpenAI GPT-5.52026.06 | 70.8 | 66.1 | 85.9 | 93.8 | 77.6 | 152 | 61 | 10 | 78 | |
| Llama-3.1-8b-instantMode=Standalone2026.06 | 70.8 | 75.2 | 56.3 | 84.8 | 79.7 | 173 | 40 | 31 | 57 | |
| Anthropic Claude Opus 4.72026.06 | 70.4 | 67 | 81.7 | 92.2 | 77.6 | 154 | 58 | 13 | 76 | |
| MedGemma-4b-it2026.06 | 64.8 | 61.3 | 76.1 | 89.2 | 72.7 | 141 | 54 | 17 | 89 |