When oncologist Dr. Sarah Lin reviewed her patient’s case in Bethesda, Maryland, she no longer had to scroll through hundreds of clinical trials by hand. Instead, she used TrialGPT, a new AI tool developed by researchers at the National Library of Medicine (NLM) and the National Cancer Institute. In seconds, the algorithm scanned through thousands of studies on ClinicalTrials.gov and returned a shortlist of trials the patient qualified for—complete with clear explanations of eligibility.
Matching patients to clinical trials has long been a slow, complex process. With over 430,000 studies listed worldwide, doctors often lack the time or tools to find the right fit. That delay can mean missed opportunities for patients and slower progress in medical research. TrialGPT changes that by using artificial intelligence to read and understand trial criteria just like a human would—but much faster.
The tool, described in a study published in Nature Communications, analyzed real-world patient cases and correctly identified relevant clinical trials with high accuracy. It doesn’t just match names or keywords—it reads the full medical profile, including diagnoses, treatments, and lab results, then explains why a patient qualifies for each trial. In one test, TrialGPT reviewed 100 patient cases and provided accurate, easy-to-understand summaries for enrollment in under five minutes total.
This isn’t just about saving time. Faster enrollment means trials can fill up quicker, accelerating the development of new treatments for cancer, rare diseases, and chronic conditions. For patients, especially those with limited options, it could mean access to cutting-edge therapies years earlier than before.
TrialGPT was created by a team led by Dr. Zhiyong Lu and Dr. Karin Verspoor at NLM’s Lister Hill National Center for Biomedical Communications. Their goal was simple: use AI not to replace doctors, but to give them superpowers. By automating the search, clinicians can focus on what matters most—talking to patients.
The tool is still in development, but its potential is already clear. With further testing and integration into hospital systems, TrialGPT could one day be as common as electronic health records. In a world where medical knowledge doubles every few years, tools like this don’t just help science keep up—they help it leap ahead.