AI Tool Tracks Clinical Trial Outcome Changes to Reduce Bias

Researchers have developed a method using a large language model (LLM) to automatically detect changes in clinical trial outcomes over time, aiming to reduce bias in medical evidence. The approach leverages historical versions of trial registrations on ClinicalTrials.gov, which archives modifications made by investigators. Traditionally, linking and comparing prespecified outcomes with those reported in publications requires extensive manual effort, limiting analyses to small numbers of trials. The new LLM-based system can process large-scale data efficiently, enabling monitoring of outcome reporting changes across thousands of trials. Presented at the 10th Peer Review Congress in September 2025, this technique could help identify selective reporting that may distort the benefits and harms of treatments.