TrialAtlas improves detection of clinical-trial design flaws in preprint tests

Published Monday, September 21, 2026 · 2:06 AM EDT

Researchers introduced TrialAtlas, a multi-agent AI system that analyzes literature, competing trials, regulatory precedents and historical drug applications to help plan clinical development. In a preprint evaluation built from 291 FDA Complete Response Letters, it achieved a 50.0% F1 score for detecting trial-design deficiencies, 6.1 points above the strongest baseline. TrialAtlas also reached 85.3% balanced accuracy and 84.7% F1 when predicting technical and regulatory success. Experts judged 86.4% of its generated concerns valid, compared with 83.1% for OpenAI DeepResearch and 59.3% for Gemini DeepResearch. The results remain benchmark and expert-evaluation findings rather than evidence from prospective clinical-trial use.

Sources

cs.CL updates on arXiv.org

In this story

OpenAI Google

Published by Tech & Business, a media brand covering technology and business. This story was sourced from cs.CL updates on arXiv.org and reviewed by the T&B editorial agent team.

Back to Newswire

Keep reading

Full wire

Science AI

EviSeek preprint reports higher accuracy with fewer tokens for embodied agents

Researchers report that EviSeek, a framework requiring no additional training, improved agents' accuracy on questions that require gathering visual evidence through movement and physical interaction.

Science AI

Bloomberg preprint reports 729 successful migrations using AI-assisted pipelines

Bloomberg researchers report 729 successful code migrations from 2,333 pull requests generated during two production campaigns, in a new preprint.

Science AI

Preprint reports AI search improves quantum error-correction codes

Researchers report in an arXiv preprint that a large language model found quantum error-correction codes that improve on published benchmarks in selected parameter regimes.

Science AI

FuseAMC preprint reports better radio signal classification on one CPU core

Researchers report in a preprint that FuseAMC, a model for identifying how radio signals encode information, reached 55.8% overall accuracy on RadioML 2016.10a while training and running on one CPU core.

AI Capital

Isomorphic Labs seeks roughly $5 billion in funding talks

Alphabet's Isomorphic Labs is in talks to raise roughly $5 billion to support its use of artificial intelligence for drug discovery, Bloomberg reported, citing people familiar with the matter. The financing remains under discussion rather than a completed fundraising.

Capital AI

AI chip startup Tenspresso seeks financing at $500 million valuation

Tenspresso is in financing talks that would value the artificial intelligence chip startup at $500 million, Bloomberg reported, citing people familiar with the matter. The three-month-old company was founded by engineers who previously worked on Google's AI chips.