Reproduce “Verifiable by Construction: How Well Do LLMs Substantiate Their Clinical Claims with Verbatim Quotes?” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Jiashuo Zhang, Yuling Chen, Yvonne Commodore-Mensah, Michael Oberst
The authors introduce an evaluation framework for checking whether clinical question-answering claims have citations, verbatim quotes from cited guideline sections, and quotes that support every point. Across 14 models, Claude Opus 5 had compliant quotes for 98.0% of claims, but only 37.1% of all claims met this strict standard. This measures verifiability, not clinical correctness.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.