The National Institutes of Health (NIH) has introduced TrialGPT, an artificial intelligence (AI) algorithm poised to transform clinical trial matching. This innovative tool analyzes patient data to swiftly identify relevant clinical trials listed on ClinicalTrials.gov, potentially accelerating medical research and enhancing trial enrollment.
A study published in Nature Communications highlights TrialGPT's capacity to accurately assess eligibility criteria and generate concise summaries for clinicians, facilitating faster and more informed decision-making. Developed collaboratively by researchers from NIH’s National Library of Medicine (NLM) and National Cancer Institute, TrialGPT leverages large language models (LLMs) to process patient summaries, exclude ineligible trials, and rank eligible studies based on relevance.
Benchmarking TrialGPT's Accuracy
To evaluate TrialGPT's accuracy, the research team benchmarked its performance against assessments made by human clinicians across more than 1,000 patient-criterion pairs. The algorithm achieved a level of accuracy nearly equivalent to that of the clinicians. Furthermore, a pilot user study revealed TrialGPT's potential to save time, with clinicians using the algorithm experiencing a 40% reduction in patient eligibility screening time while maintaining accuracy levels comparable to manual methods. This efficiency could free up healthcare providers to concentrate on more complex tasks requiring human expertise.
Addressing Barriers to Clinical Research
Clinical trials are essential for advancing medical discoveries, but the process of identifying eligible participants is often labor-intensive and can delay progress. TrialGPT is designed to mitigate these challenges, potentially increasing access to trials for underrepresented populations and improving overall recruitment effectiveness.
"Our study shows that TrialGPT could help clinicians connect their patients to clinical trial opportunities more efficiently and save precious time that can be better spent on harder tasks that require human expertise," said Zhiyong Lu, NLM Senior Investigator and corresponding author of the study.
Future Evaluation
In recognition of its promising results, the TrialGPT research team has been awarded The Director’s Challenge Innovation Award. This award will support further evaluation of the model’s real-world performance and ensure its fairness across diverse populations. The study was a collaborative effort involving researchers from Albert Einstein College of Medicine, the University of Pittsburgh, the University of Illinois Urbana-Champaign, and the University of Maryland, College Park.
