Researchers at the National Institutes of Health (NIH) have developed TrialGPT, an AI algorithm that streamlines the process of matching potential volunteers to clinical trials on ClinicalTrials.gov. A study in Nature Communications found TrialGPT can accurately identify relevant trials, explain how a person meets eligibility criteria, and provide an annotated list ranked by relevance, enabling clinicians to discuss options with patients more efficiently.
Developed by NIH’s National Library of Medicine and National Cancer Institute, TrialGPT uses large language models (LLMs) to analyze patient summaries, filter eligible trials, and exclude ineligible ones. In tests comparing its accuracy to human clinicians, TrialGPT performed nearly as well, while clinicians using the tool spent 40% less time screening patients.
Connecting patients to clinical trials is crucial for advancing medical research, yet the process is often slow and resource-intensive. TrialGPT has the potential to improve trial enrollment, accelerate discoveries, and reduce participation barriers for underrepresented populations. The research team, recently awarded The Director's Challenge Innovation Award, plans further testing to ensure fairness and efficacy in real-world clinical settings.
Click here to read the original news story.