"I Know Where to Look," But Does the LLM? Charting the Gaps Between Clinical Expert Needs and Unstructured Data Abstraction Tools
3.40T1 sourcearXiv cs.HC
arXiv cs.HCoriginal source ↗
Source record
Published by arXiv cs.HC (T1 source). The original is at https://arxiv.org/abs/2609.19318.
Pipeline notes
The summary and note below are generated by the signal pipeline — they are Beyond Desk’s reading, not quotations from the source.
SummaryCo-design study with seven cancer research teams using an LLM-based clinical data abstraction system called Libretto found clinicians struggled to replicate their annotation intuitions with LLMs in 12 of 14 tasks, due to note reliability judgments, prompt steering difficulties, and inflexible evaluation strategies.
Why it mattersNames concrete friction between domain experts and LLM extraction tools, useful for anyone building or evaluating similar systems beyond template demos.
Cited by
No citations on record.