New RECITE framework uses LLMs to revise clinical trial eligibility criteria

Researchers have developed RECITE, a framework designed to automate the process of discovering, extracting, and implementing evidence-based revisions for clinical trial eligibility criteria. This system utilizes protocol amendment documents to create a dataset of 11,913 instances, pairing original criteria with supporting evidence and proposed changes. Benchmarking various LLM configurations, the study found that Gemma 2 9B and GPT-4o performed comparably, achieving high scores in an LLM-as-judge protocol validated by domain experts. The framework also identified 44 relevant paper-trial matches through an agentic literature discovery system, suggesting that existing clinical knowledge can be operationalized to improve trial design. AI

IMPACT This framework could streamline clinical trial design by leveraging LLMs to analyze and update eligibility criteria, potentially accelerating research and patient enrollment.

RANK_REASON The item is an academic paper detailing a new framework and its evaluation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• alphaXiv
• arXiv
• CatalyzeX Code Finder for Papers
• Connected Papers
• CORE Recommender
• DagsHub
• Gemma 2 9B
• Gotit.pub
• GPT-4o
• Hugging Face
• Litmaps
• ScienceCast
• scite Smart Citations

• paper
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →