GPT-4.1 with RAG hits 92.1% accuracy reviewing clinical trial informed consent forms
lamechthinkbig · x · 2026-10-19
Researchers at Boehringer Ingelheim published a JMIR Medical Informatics paper evaluating GPT-4.1 with retrieval-augmented generation (RAG) for extracting data-reuse information from multilingual clinical trial informed consent forms (ICFs).
- Initial run on 438 ICFs across countries, languages, and versions: 81.6% accuracy;
- Validation on 488 ICFs with 3 iterative optimization rounds: baseline 84.0%, improving to 83.2%, 90.7%, then 92.1% on an independent 104-document holdout cohort;
- The system accurately extracted data-reuse language, supporting ICF review and improving consistency.
Conclusion: RAG mitigates hallucinations, making LLM-assisted ICF review feasible in high-stakes settings.
More from Safety
- Microsoft's ASSERT and ACS: blocking AI agents from accessing the wrong account — usamawahabkhan · 2026-10-19
- OpenRouter details Sentinel, a self-improving 24/7 security agent with 0.05% false-positive rate — Dan_Jeffries1 · 2026-10-19
- One refusals at a time: user says Claude's automated safety checks make the product unusable — DubaiInJuly · 2026-10-19
- Claude's answer to Musk's Grok-DMs tip nails the prompt injection risk — alex_verem · 2026-10-19
- Musk warns many will be 'one-shotted' by super-persuasive superintelligences — elonmusk · 2026-10-19
- Gemini's critical summary of a safety report surfaces downplayed residual risks — davidmanheim · 2026-10-19