Clinical Value Set Authoring on RASC full (test)
62.5PrecisionExpanded pool + GPT-5 adjudication
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Expanded pool + GPT-5 adjudicationEvaluation set=full test2026.06 | 62.5 | 61.3 | 54.9 | |
| RASC SAPBert cross-encoderEvaluation set=full test2026.06 | 27.2 | 48.3 | 29.8 | |
| Expanded pool + SAPBert cross-encoderEvaluation set=full test2026.06 | 26.9 | 49.9 | 28.7 | |
| GPT-4o zero-shot generationEvaluation set=full test2026.06 | 16.9 | 10 | 10.5 | |
| RASC retrieval-onlyEvaluation set=full test2026.06 | 9.2 | 55.3 | 13.6 |