AI-Driven Chart Review Accurately Identifies Potential Rare Disease Trial Participants in New Study

A study by Cleveland Clinic and Dyania Health shows that a medically trained large language model can screen electronic medical records to identify patients eligible for a rare disease clinical trial. The research was published in The Journal of Cardiac Failure. It assessed an AI system developed by Dyania Health and deployed at Cleveland Clinic for pre-screening participants in the DepleTTR-CM Phase 3 trial for transthyretin amyloid cardiomyopathy. In one week the system reviewed 1,476 patients and identified 46 as potential matches. Of the 30 patients flagged by the AI, 36.6 percent were Black, compared with 7.1 percent identified through routine screening. Only 60 percent of the AI-identified patients had prior connections to a heart failure specialist, versus 92.8 percent of those found by traditional methods. The AI platform, called Synapsis AI, combined structured electronic medical record data with natural language processing of clinical notes and lab reports. It generated auditable justifications for each inclusion or exclusion decision. The system screened data across 25 hospitals and 250 outpatient centers in Ohio, Florida and Nevada, with clinical team validation required as part of the workflow. Trejeeve Martyn, lead study investigator and director of Heart Failure Population Health at Cleveland Clinic, said the technology supports chart review at scale and can increase enrollment of patients from different backgrounds. Eirini Schlosser, chief executive of Dyania Health, said the approach helps surface eligible patients who may otherwise be missed. Cleveland Clinic has invested in Dyania Health.