Are you a cancer researcher or data scientist interested in applying artificial intelligence (AI) and large language models (LLMs) to oncology? Learn how a team of researchers from NIH’s National Library of Medicine (NLM) and NCI developed “TrialGPT”—a new AI algorithm making it easier for both clinicians and patients to find and connect with the right clinical trial opportunities. Read on for both a summary of what this team discovered and a GitHub link to TrialGPT’s code and data.

As you know, matching patients to suitable clinical trials has been a complex and time-consuming process until now. LLMs’ advanced language understanding and contextual reasoning capabilities make it possible to overcome data and transparency limitations in the matching process.

This pilot user study that the researchers conducted demonstrates that TrialGPT significantly reduces the screening time needed for human experts. This study and algorithm pave the way for valuable LLMs to assist in the process of patient-trial matching.

“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 NLM Senior Investigator and corresponding author of the study, Zhiyong Lu, Ph.D.

NCI researcher, Dr. Floudas, shared, “Our pilot user study mimicked the actual clinical matching task at NCI, and results show that TrialGPT can reduce patient screening time by 42.6%, offering improved efficiency for clinical trial recruitment.”

NCI study co-authors, Drs. Xue and Bracken-Clarke also added, “Unlike other pilot studies that explored using LLMs to enhance first-stage retrieval of clinical trials, our study presents an end-to-end solution (trial retrieval, matching, and ranking) to streamline clinical trial recruitment with LLMs.”