In a groundbreaking advancement at the nexus of artificial intelligence and cardiology, Cleveland Clinic and Dyania Health have unveiled promising research that demonstrates how an AI-powered, medically trained large language model system can revolutionize the identification process for clinical trial candidates. This innovation is shown to efficiently and accurately sift through electronic medical records (EMRs), pinpointing eligible patients for rare disease clinical trials with unprecedented precision and speed.
Published in the prestigious Journal of Cardiac Failure, the study highlights a leap forward in the operational efficiency of medical chart reviews. It uncovers that AI can significantly enhance not just the velocity but also the accuracy and equitable inclusion criteria for trial enrollment, ensuring a broader and more diverse patient demographic is considered for participation. This is particularly relevant in the context of heart failure subtypes, such as transthyretin amyloid cardiomyopathy (ATTR-CM), a condition predominantly affecting elderly populations and typically challenging to detect through traditional means.
The AI solution, developed by Dyania Health and deployed across Cleveland Clinic’s vast network, was tasked with pre-screening patients for the DepleTTR-CM Phase 3 clinical trial. This system thoughtfully merges structured EMR data with cutting-edge natural language processing, parsing through complex clinical narratives, lab reports, and unstructured notes. Over just one week, it reviewed a staggering 1,476 individual patient records and flagged 46 individuals as potential participants, underscoring the power of AI to scale chart review operations in a manner unachievable by human effort alone.
Equally compelling is the system’s clinical accuracy. The AI model demonstrated a 96.2% precision rate when it answered approximately 7,700 trial-specific questions across nine distinct clinical domains, thereby maintaining rigorous compliance with the trial’s inclusion and exclusion criteria. Moreover, it provided fully auditable and physician-interpretable justifications for each eligibility decision, a feature that enhances transparency and builds clinician trust in AI-driven methodologies.
One of the most vital outcomes cited in the study is the AI’s negative predictive value (NPV), hitting an impressive 99% through correctly excluding 198 out of 200 non-eligible patients. This high NPV not only minimizes unnecessary follow-ups but also streamlines the workflow for clinical research teams, allowing them to concentrate on high-potential candidates with confidence in the system’s reliability.
Perhaps one of the most transformative findings is the impact of AI on increasing diversity and equity in clinical trial recruitment. Out of the 30 patients accurately identified by AI, 36.6% were Black, a stark contrast to the mere 7.1% identified through standard screening processes. This suggests that AI systems are uniquely positioned to discover eligible patients from traditionally underrepresented groups, thereby addressing long-standing disparities in clinical research participation.
Furthermore, the study revealed that only 60% of AI-identified patients had prior connections to heart failure specialists, compared to 92.8% in the traditionally detected group. This underscores the AI’s capacity to access patient populations that might otherwise be overlooked, expanding the reach of clinical trials beyond existing specialist networks and potentially uncovering untapped pools of patients who could benefit from new therapies.
Dr. Trejeeve Martyn, lead investigator and director of Heart Failure Population Health at Cleveland Clinic, emphasizes that this technology marks a paradigm shift in clinical trial recruitment practices. By automating chart review at scale, AI frees research teams from the traditionally labor-intensive and time-consuming manual processes, accelerating enrollment and enabling trials to meet—and potentially exceed—target goals more rapidly.
The AI model’s integration within Cleveland Clinic’s EMR infrastructure spans an extensive network, including 25 hospitals and 250 outpatient centers across Ohio, Florida, and Nevada. This wide deployment demonstrates the system’s scalability and adaptability to diverse healthcare settings, making it a viable tool for extensive population health management and real-world clinical trial orchestration.
Dyania Health’s Synapsis AI combines domain-specific machine learning techniques with natural language understanding to dissect the nuanced language found within clinical notes. This hybrid approach facilitates a deeper understanding of patient records, lending itself to more accurate abstraction of complex data points which historically required intensive manual chart reviews.
Despite the high degree of automation, the workflow preserves essential clinical oversight. Human validation remains a core part of the process, safeguarding patient safety and ensuring the AI’s decisions align with medical standards. This clinician-in-the-loop model exemplifies how AI and human expertise can synergize, bringing efficiency without compromising the rigor of clinical trial enrollment protocols.
Industry experts see this as a harbinger of broader applications for AI in healthcare beyond trial matching. The technology holds promise for accelerating observational research studies, bolstering disease registries, and facilitating the implementation of evidence-based, yet underutilized, treatments across healthcare systems. Improved data abstraction capabilities and real-time quality reporting could transform how medical institutions monitor and improve patient outcomes.
Echoing these sentiments, Eirini Schlosser, CEO and Co-founder of Dyania Health, highlights the bottleneck clinical research faces due to inefficient, manual patient matching systems. The Synapsis AI platform addresses this challenge head-on, potentially reshaping recruitment workflows while promoting inclusivity and access to cutting-edge clinical trials for patients historically marginalized in medical research.
Financially intertwined with the technological advancement, Cleveland Clinic has invested in Dyania Health and stands to benefit from the commercialization of this AI-driven chart review innovation. This strategic partnership exemplifies a successful model of healthcare institutions fostering AI startups to bridge technological innovation with clinical impact.
As the healthcare industry grapples with burgeoning data volumes and the pressing need for rapid, equitable access to clinical trials, this study from Cleveland Clinic and Dyania Health illuminates a path forward. The amalgamation of AI and clinical expertise not only optimizes the recruitment process but also addresses persistent challenges in patient diversity, underscoring AI’s pivotal role in the future of precision medicine and clinical research.
Subject of Research:
Artificial intelligence-enabled medical chart review for clinical trial eligibility in transthyretin amyloid cardiomyopathy (ATTR-CM).
Article Title:
Automating Chart Review Utilizing an Artificial Intelligence-Enabled System for Assessing Transthyretin Amyloid Cardiomyopathy Trial Eligibility.
News Publication Date:
March 3, 2026.
Keywords:
Cardiovascular disorders, Cardiomyopathy, Heart failure, Cardiology, Artificial intelligence, Computer science, Clinical studies.
