Ro, Russell, Abbasi-Asl, Reza · Oct 30, 2026
Russell Ro and Reza Abbasi-Asl introduce RECITE for clinical-trial criteria
Markdown · JSON · Original source
Russell Ro and Reza Abbasi-Asl introduce RECITE, a framework for finding and applying evidence-based revisions to clinical-trial eligibility criteria. They describe a gap: overly restrictive criteria contribute to enrollment failure, while published recommendations for broadening criteria remain disconnected from trial design at scale. RECITE uses protocol amendment documents to pair original criteria with documentary evidence and target revisions. The authors report a dataset spanning thousands of trials and fifteen configurations using an LLM-as-judge protocol validated against ten domain experts. The best reported and closed-weight models receive similar ordinal scores on that benchmark; these scores do not demonstrate performance in changing trial enrollment. The authors also report 44 paper–trial matches across five publications, identified using an literature discovery system. Their argument is that existing clinical knowledge can be operationalized into actionable trial-design recommendations. The evidence may help connect proposed criteria changes to documentary precedents, but the abstract does not establish whether applying RECITE recommendations reduces enrollment failure. The reported assessment also relies on an LLM judge, whose agreement with human experts is reported but is not perfect.
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Original source
cs.AI · Article
Source details
Original source date: Oct 30, 2026 (2026-10-30).
Date basis: Date basis not provided.
Added to Escape Meridian: 2026-11-11T01:22:43.172Z. First observed: 2026-11-02T05:04:08.245Z. Retrieved: 2026-11-02T07:49:29.331Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
The abstract reports scores under an -as-judge protocol, but does not report whether applying RECITE recommendations in trial design reduces enrollment failure. The reported judge–human agreement is not perfect: the abstract gives quadratic-weighted Cohen's κ of 0.694 ± 0.096. The supplied evidence is the paper's abstract; it does not include the full methods or results needed to assess the benchmark and reported paper–trial matches in detail.
Open question
Would applying RECITE’s evidence-based eligibility-criteria revisions in trial design reduce enrollment failure?
Source-stated
The authors say overly restrictive eligibility criteria contribute to widespread enrollment failure, while recommendations for broadening criteria are disconnected from trial design at scale.
Overly restrictive eligibility criteria contribute to widespread enrollment failure in clinical trials, yet published recommendations for how specific criteria should be broadened remain disconnected from trial design at scale.
Source-stated
RECITE uses protocol amendment documents to create a dataset pairing original eligibility criteria with documentary evidence and target revisions.
construct a dataset of 11,913 curated instances across 5,735 trials, each pairing original criteria with documentary evidence and target revisions.
Source-stated
The authors benchmarked fifteen LLM configurations and report the best open-weight and closed-weight model results under their LLM-as-judge protocol.
The best open-weight model, Gemma 2 9B, achieves $\kappa_{qw}$-aligned ordinal score $3.0 \pm 0.9$ on a 0--4 scale (normalized 0.742) and the best closed-weight model, GPT-4o, reaches $3.0 \pm 0.8$ (normalized 0.745).
Source-stated
The authors report that the judge protocol was validated against ten domain experts, with a reported agreement score.
under an LLM-as-judge protocol validated against ten domain experts (judge vs. human quadratic-weighted Cohen's $\kappa_{qw} = 0.694 \pm 0.096$).
Source-stated
The authors say their literature discovery system identified paper–trial matches that provide trial-design and eligibility-criteria-modification evidence.
Using an agentic literature discovery system, we then identified 44 paper--trial matches across five publications providing trial-design and EC-modification evidence.
Source-stated
The authors conclude that existing clinical knowledge may be operationalized as actionable trial-design recommendations.
Our findings suggest that existing clinical knowledge can be operationalized into actionable trial design recommendations.
Source and contributor context
Ro, Russell, Abbasi-Asl, Reza · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
No suitable media was found by the supported page extraction.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Benchmark
A defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Large language model
A language model with many learned numerical settings, used to process and generate language.
Open weights
A model whose trained numbers are published so others can download, run and inspect it; the license and training data may still be restricted.
Agentic AI
AI systems that plan and carry out a series of steps toward a goal, often by using other software, rather than giving one answer.
arXiv
arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Editorial provenance
Version 1. Publication ID: 8f2df142-1cc1-4cfa-88b6-47e2c974f362.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: passed by anthropic/claude-opus-5.5 on 2026-11-11T01:22:43.172Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: material_development.
Observation: abaf63faf86d22e4cb89747ba38a377a768e00728819cb77ce4afb04ebf0c1d1. Source text hash: 96a04428986878e0455d2eed727c93db80b7e6efaf153701c50d1adaf58b09fc.
Revision and source-check history
Version 1 · 2026-11-11T01:22:43.172Z
Russell Ro and Reza Abbasi-Asl introduce RECITE for clinical-trial criteria
Russell Ro and Reza Abbasi-Asl introduce RECITE, a framework for finding and applying evidence-based revisions to clinical-trial eligibility criteria. They describe a gap: overly restrictive criteria contribute to enrollment failure, while published recommendations for broadening criteria remain disconnected from trial design at scale. RECITE uses protocol amendment documents to pair original criteria with documentary evidence and target revisions. The authors report a dataset spanning thousands of trials and fifteen configurations using an LLM-as-judge protocol validated against ten domain experts. The best reported and closed-weight models receive similar ordinal scores on that benchmark; these scores do not demonstrate performance in changing trial enrollment. The authors also report 44 paper–trial matches across five publications, identified using an literature discovery system. Their argument is that existing clinical knowledge can be operationalized into actionable trial-design recommendations. The evidence may help connect proposed criteria changes to documentary precedents, but the abstract does not establish whether applying RECITE recommendations reduces enrollment failure. The reported assessment also relies on an LLM judge, whose agreement with human experts is reported but is not perfect.
Original source
cs.AI · Article
Source details
Original source date: Oct 30, 2026 (2026-10-30).
Date basis: Date basis not provided.
Added to Escape Meridian: 2026-11-11T01:22:43.172Z. First observed: 2026-11-02T05:04:08.245Z. Retrieved: 2026-11-02T07:49:29.331Z.
Limits
The abstract reports scores under an -as-judge protocol, but does not report whether applying RECITE recommendations in trial design reduces enrollment failure. The reported judge–human agreement is not perfect: the abstract gives quadratic-weighted Cohen's κ of 0.694 ± 0.096. The supplied evidence is the paper's abstract; it does not include the full methods or results needed to assess the benchmark and reported paper–trial matches in detail.
Open question
Would applying RECITE’s evidence-based eligibility-criteria revisions in trial design reduce enrollment failure?
Source-stated
The authors say overly restrictive eligibility criteria contribute to widespread enrollment failure, while recommendations for broadening criteria are disconnected from trial design at scale.
Overly restrictive eligibility criteria contribute to widespread enrollment failure in clinical trials, yet published recommendations for how specific criteria should be broadened remain disconnected from trial design at scale.
Source-stated
RECITE uses protocol amendment documents to create a dataset pairing original eligibility criteria with documentary evidence and target revisions.
construct a dataset of 11,913 curated instances across 5,735 trials, each pairing original criteria with documentary evidence and target revisions.
Source-stated
The authors benchmarked fifteen LLM configurations and report the best open-weight and closed-weight model results under their LLM-as-judge protocol.
The best open-weight model, Gemma 2 9B, achieves $\kappa_{qw}$-aligned ordinal score $3.0 \pm 0.9$ on a 0--4 scale (normalized 0.742) and the best closed-weight model, GPT-4o, reaches $3.0 \pm 0.8$ (normalized 0.745).
Source-stated
The authors report that the judge protocol was validated against ten domain experts, with a reported agreement score.
under an LLM-as-judge protocol validated against ten domain experts (judge vs. human quadratic-weighted Cohen's $\kappa_{qw} = 0.694 \pm 0.096$).
Source-stated
The authors say their literature discovery system identified paper–trial matches that provide trial-design and eligibility-criteria-modification evidence.
Using an agentic literature discovery system, we then identified 44 paper--trial matches across five publications providing trial-design and EC-modification evidence.
Source-stated
The authors conclude that existing clinical knowledge may be operationalized as actionable trial-design recommendations.
Our findings suggest that existing clinical knowledge can be operationalized into actionable trial design recommendations.
Source and contributor context
Ro, Russell, Abbasi-Asl, Reza · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
No suitable media was found by the supported page extraction.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Benchmark
A defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Large language model
A language model with many learned numerical settings, used to process and generate language.
Open weights
A model whose trained numbers are published so others can download, run and inspect it; the license and training data may still be restricted.
Agentic AI
AI systems that plan and carry out a series of steps toward a goal, often by using other software, rather than giving one answer.
arXiv
arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Editorial provenance
Version 1. Publication ID: 8f2df142-1cc1-4cfa-88b6-47e2c974f362.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: passed by anthropic/claude-opus-5.5 on 2026-11-11T01:22:43.172Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: material_development.
Observation: abaf63faf86d22e4cb89747ba38a377a768e00728819cb77ce4afb04ebf0c1d1. Source text hash: 96a04428986878e0455d2eed727c93db80b7e6efaf153701c50d1adaf58b09fc.
passed · 2026-11-11T01:22:43.172Z · anthropic/claude-opus-5.5
Publication: 8f2df142-1cc1-4cfa-88b6-47e2c974f362
attribution — pass: The reading correctly attributes RECITE and its claims to Russell Ro and Reza Abbasi-Asl, the listed authors. It frames the findings and conclusions as the authors' reports and arguments rather than as established facts.
Authors:Russell Ro, Reza Abbasi-Asl
summary — pass: The title, summary and question faithfully capture the abstract. It explains the problem (restrictive criteria and enrollment failure), the dataset, the benchmark, the literature matches and the authors' conclusion. The question about enrollment impact follows from an unresolved issue. Terms such as LLM-as-judge, open-weight and ordinal score are kept exact and can be explained through tooltips. They do not prevent a general reader from understanding the central development.
a framework for automated discovery, extraction, and implementation of evidence-based eligibility criteria revisions
date — pass: The originalPublished date of 2026-10-30 matches the arXiv v1 submission date. The later feed date of November 2 is not used as the publication date.
[Submitted on 30 Oct 2026]
support — pass: Each claim is directly supported by abstract text. The reading avoids turning benchmark scores into real-world reliability and explicitly notes that enrollment impact is not shown. It also notes that judge agreement with experts is imperfect. Its one inference, about connecting criteria changes to precedents, is conditional and grounded in the dataset's pairing of criteria with documentary evidence.
each pairing original criteria with documentary evidence and target revisions
media — pass: The selected media is the verified direct arXiv abstract page for this paper. It is the most direct access point to the source.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
A defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Definition referenceA language model with many learned numerical settings, used to process and generate language.
Definition referenceA model whose trained numbers are published so others can download, run and inspect it; the license and training data may still be restricted.
Definition referenceAI systems that plan and carry out a series of steps toward a goal, often by using other software, rather than giving one answer.
Definition referencearXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Explore arXivA defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Definition referenceA language model with many learned numerical settings, used to process and generate language.
Definition referenceA defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Definition referenceA language model with many learned numerical settings, used to process and generate language.
Definition referenceA model whose trained numbers are published so others can download, run and inspect it; the license and training data may still be restricted.
Definition referenceAI systems that plan and carry out a series of steps toward a goal, often by using other software, rather than giving one answer.
Definition referencearXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Explore arXivA defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Definition referenceA language model with many learned numerical settings, used to process and generate language.
Definition reference