A new study by researchers at Mass General Brigham highlights how artificial intelligence (AI) can significantly speed up patient screening for clinical trial enrollment.
Image Credit: New Africa/Shutterstock.com
Their AI-assisted screening tool drastically improved the efficiency of determining eligibility and enrolling patients in a heart failure clinical trial compared to traditional manual methods. The findings suggest that AI-driven screening can reduce costs and accelerate research, potentially giving patients earlier access to effective treatments.
Seeing this AI capability accelerate screening and trial enrollment this substantially in the context of a real-world randomized prospective trial is exciting. We look forward to using this capability to assist as many trials as we can.
Samuel (Sandy) Aronson, ALM, MA, Study Co-Senior Author and Executive Director, IT and AI Solutions, Mass General Brigham Personalized Medicine
The study included 4476 patients, who were randomly assigned to either a manual screening process or an AI-assisted screening using a generative AI tool called RAG-Enabled Clinical Trial Infrastructure for Inclusion Exclusion Review (RECTIFIER). The AI analyzed clinical notes and electronic health records to assess whether patients met key eligibility criteria, such as symptoms, chronic conditions, and medication history. Research staff then conducted a brief review of AI-identified charts to confirm eligibility.
In the manual screening group, study staff reviewed patient charts without AI assistance to determine eligibility. Over a set period, the AI-assisted process screened 458 eligible patients, compared to 284 in the manual group—a considerable improvement in efficiency.
Once patients were identified as eligible, navigators reached out to gauge their willingness to participate in the trial. These navigators were blinded to whether the screening was AI- or human-based to prevent bias. In the AI group, 35 patients enrolled in the trial, nearly double the 19 who enrolled in the manual group.
The rate of enrollment in the AI-enabled arm was almost double the rate of enrollment in the manual arm. This means that AI could almost halve the time it takes to complete enrollment in a trial.
Ozan Unlu, MD, Study Lead Author and Fellow, Clinical Informatics, Mass General Brigham, Cardiovascular Medicine, Brigham and Women's Hospital
Given concerns about potential bias in AI models, the researchers analyzed race, gender, and ethnicity among enrolled patients and found no significant differences between those screened manually and those screened with AI assistance.
This study builds on earlier research by Blood, Aronson, Unlu, and colleagues, published in NEJM AI in June, which demonstrated that RECTIFIER was slightly more accurate than manual screening in identifying eligible patients in a retrospective review. The latest findings validate its effectiveness in real-world clinical settings.
Our next goal is to expand the AI screening tool’s use outside of Mass General Brigham. By adjusting the eligibility questions that the RECTIFIER tool asks of the medical record notes, AI screening can be applied to trials assessing cancer treatments, diabetes interventions, and many others.
Alexander Blood, MD, MSc, Study Co-Senior Author and Cardiologist, Brigham and Women’s Hospital
Journal Reference:
Unlu, O., et al. (2025). Manual vs AI-Assisted Prescreening for Trial Eligibility Using Large Language Models—A Randomized Clinical Trial. JAMA. doi.org/10.1001/jama.2024.28047.