AI-Driven Chart Review Improves Identification of ATTR-CM Trial Candidates

Artificial intelligence (AI) can significantly improve the process of identifying patients eligible for rare disease clinical trials, according to a new study conducted by researchers at Cleveland Clinic and Dyania Health.¹

Using a medically trained large language model embedded within the electronic health record (EHR) system, the researchers were able to screen more than 1400 patient records in 1 week and accurately identify potential candidates for a phase 3 trial targeting transthyretin amyloid cardiomyopathy (ATTR-CM), a form of heart failure that primarily affects older adults.

“This study shows how medically trained AI can support chart review at scale, transforming what has traditionally been a labor-intensive process,” said Trejeeve Martyn, MD, lead study investigator and director of Heart Failure Population Health at Cleveland Clinic, in a statement. “By rapidly identifying high-quality trial candidates across a large health system, we can increase enrollment efficiency and increase enrollment of patients from different backgrounds and from a broader geographical area. We are optimistic that this technology can be used across our health system and are looking at how the platform can help accelerate observational research, disease registries and evidence-based implementations of approved therapies that are underutilized.”

ATTR-CM is a progressive and often underdiagnosed heart condition caused by the buildup of misfolded transthyretin proteins in the heart muscle.² These abnormal protein deposits, known as amyloid, accumulate in the heart walls, making them stiff and impairing the heart’s ability to relax and pump blood effectively, which can lead to heart failure.

The model used a combination of structured EHR data and natural language processing to analyze both discrete clinical variables and unstructured information from physician notes, laboratory results, and reports.³ The AI system reviewed 1476 patient records across the health system and evaluated trial eligibility by answering approximately 7700 trial-specific questions spanning multiple inclusion and exclusion domains. Its determinations were accompanied by auditable justifications for each decision, which were subsequently verified by clinicians in a “clinician-in-the-loop” review process to confirm eligibility and ensure accuracy before patient outreach.

Of the 30 patients ultimately confirmed as eligible through the AI system and clinician-in-the-loop review, 29 had not been identified through traditional recruitment approaches.¹ The AI achieved an overall accuracy of 96.2%. The system’s explanations for its eligibility determinations were also fully interpretable, with physician reviewers confirming 100% accuracy in the AI-generated justifications.

In terms of recruitment outcomes, 7 patients were enrolled through AI-assisted screening before the site met its enrollment goal, compared with 10 patients identified through conventional screening over a 90-day period. Additionally, the system demonstrated strong exclusion capability, correctly ruling out 198 of 200 noneligible patients and achieving a 99% negative predictive value.

Importantly, the AI-driven screening process identified a more diverse group of potential trial participants. Among the 30 patients identified through the AI system, 36.6% were Black, compared with just 7.1% of those identified through routine recruitment methods. The AI approach also reached patients who were less connected to specialty care: Only 60% of AI-identified patients had previously seen a heart failure specialist, compared with 92.8% of those identified through traditional screening.

These findings suggest that AI-assisted chart review may help expand access to clinical trials by identifying eligible patients who might otherwise be overlooked in conventional recruitment pathways, particularly among historically underenrolled populations.

“Clinical research is often limited by how efficiently and equitably we can match patients to trials,” said Eirini Schlosser, CEO and cofounder of Dyania Health, in a statement. “This study provides compelling evidence that AI can help solve that bottleneck—not just by improving workflow efficiency, but by helping surface eligible patients who may otherwise be missed, especially those from historically underrepresented groups.”

References

1. AI-driven chart review accurately identifies potential rare disease trial participants in new study. Cleveland Clinic. March 3, 2026. Accessed March 5, 2026. https://www.eurekalert.org/news-releases/1118354

2. Transthyretin amyloid cardiomyopathy (ATTR-CM). American Heart Association. Last reviewed May 29, 2024. Accessed March 5, 2026. https://www.heart.org/en/health-topics/cardiomyopathy/what-is-cardiomyopathy-in-adults/transthyretin-amyloid-cardiomyopathy-attr-cm

3. Martyn T, Hanna M, Nissen SE, et al. Automating chart review utilizing an artificial intelligence-enabled system for assessing transthyretin amyloid cardiomyopathy trial eligibility. J Card Fail. Published online March 3, 2026. doi:10.1016/j.cardfail.2026.01.010