Researchers from the National Institute of Health( NIH) developed an Artificial Intelligence called TrialGPT to effectively navigate the process of matching potential volunteers to relevant research trials.

Recruiting participants for clinical trials has always been a challenging task. TrialGPT, a cutting-edge framework leveraging large language models, promises to make this process seamless and efficient. This innovative system operates through three modules—Retrieval identifies potential trials, Matching evaluates patient eligibility, and Ranking prioritizes trials with a scoring system.

In tests with synthetic patient data and trial annotations, TrialGPT achieved 87.3% accuracy in eligibility predictions and significantly outperformed existing models in trial ranking. It also reduced screening time by 42.6%, showcasing its practical utility.

By recalling over 90% of relevant trials and providing explanations close to expert performance, TrialGPT paves the way for a more effective and human-centered approach to patient-to-trial matching.