LLM Filtering on Manually adjudicated gold-standard CUIs Chronic Heart Failure v1 (test)
137CUIs CountGPT-5-mini
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| GPT-5-miniFramework=CUICurate, Backbone=GPT-5-mini2026.02 | 137 | 92 | 87 | 89 | |
| GPT-5Framework=CUICurate, Backbone=GPT-52026.02 | 119 | 86 | 93 | 90 | |
| manualMode=manual curation2026.02 | 98 | 74 | 97 | 84 |