LLM Filtering on Manually adjudicated gold-standard CUIs Poor Mobility v1 (test)
205CUIs CountGPT-5-mini
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| GPT-5-miniFramework=CUICurate, Backbone=GPT-5-mini2026.02 | 205 | 86 | 87 | 86 | |
| GPT-5Framework=CUICurate, Backbone=GPT-52026.02 | 171 | 76 | 93 | 83 | |
| manualMode=manual curation2026.02 | 92 | 23 | 52 | 32 |