CLINICAL QUESTION: This study compares the efficiency of an AI-assisted screening tool versus manual chart reviews for unstructured data in assessing patient eligibility based on specific criteria.
BACKGROUND: Eligibility-based patient recruitment for clinical trials is time-consuming, labor-intensive, and expensive. Structured electronic health records helped improve this process; however, they still need a manual chart review for unstructured data. The authors developed a large language model tool, Retrieval Augmented Generation Enabled Clinical Trial Infrastructure for Inclusion Exclusion Review (RECTIFIER), that parses unstructured data.
STUDY DESIGN: Single-center, prospective, blind, randomized clinical trial
SETTING: Mass General Brigham Health System
SYNOPSIS: Study included 4,476 patients based on structured criteria between May 31 to Sept 28, 2024. They were randomized to two groups: manual screening by study staff or AI-assisted screening. RECTIFIER screening identified eligible patients significantly faster (hazard ratio, 1.78; P <.001). They found a higher eligibility rate, 20.4% (458/2,242 patients) in the AI group versus 12.7% (284/2,234 patients) in the manual group (P <.001). In the end, 35 patients were enrolled via AI versus 19 via manual screening (P = .04). Using cumulative incidence of eligibility determination and enrollment, the proportion of eligible patients was similar between the groups (20.8% [458/2,205] for the AI screening group and 21.1% [284/1,347] for the manual screening group; P = .86). More than 99% of AI-screened patients were processed within 15 days, compared to 50 days for manual screening.
Limitations include a single-center study focused on heart failure. Hence, this would need a broader validation across several sites and other diagnoses. Despite the limitations, the study shows AI-assisted screening significantly improved trial screening speed and enrollment.
BOTTOM LINE: AI-assisted technology implementation for screening is a promising tool for accelerating clinical research and reducing costs.
CITATION: Unlu O, et al. Manual vs AI-assisted prescreening for trial eligibility using large language models—a randomized clinical trial. JAMA. 2025;333(12):1084-1087. doi: 10.1001/jama.2024.28047.
Dr. Jain
Dr. Kenney
Dr. Jain is a hospitalist in the department of internal medicine at The Ohio State University Wexner Medical Center and an assistant professor of medicine at The Ohio State University College of Medicine, both in Columbus, Ohio. Dr. Kenney is a hospitalist at The Ohio State University Wexner Medical Center and an assistant professor of medicine at The Ohio State University College of Medicine, both in Columbus, Ohio.