Unmet Need
Hypertrophic cardiomyopathy (HCM) is a rare genetic condition (prevalence in adults: 0.2-0.5%) that causes pathologic enlargement of the left ventricle (1). While many HCM patients live normal lives, some are at risk for severe cardiac complications, including sudden cardiac death due to ventricular arrhythmia (VA). VA risk is currently evaluated in HCM patients based on standardized clinical criteria published by the American College of Cardiology (ACC), American Heart Association (AHA), and the European Society of Cardiology (ESC). These guidelines have been reported to have inconsistent accuracies in different cohorts, with lower diagnostic capabilities for persons of color and ethnic minority groups. In addition, they often fail to incorporate raw data from advanced cardiac imaging, which may be helpful for risk stratification. Therefore, there is a strong need to develop methods that more accurately assess VA risk to improve health outcomes for HCM patients.
Value Proposition
· Enhanced accuracy for predicting ventricular arrhythmia in HCM patients
· Improved predictive capability for young patients, persons of color, and ethnic minorities
· Incorporates analysis of raw cardiac MRI data, which may capture risk factors not accounted for in standard radiology reports
· Explainable and more transparent prediction by providing insights into how a patient’s data contribute to their final arrhythmia risk
Technology Description
· Researchers at Johns Hopkins have developed a deep learning method for determining VA risk in HCM patients.
· Capable of incorporating clinical risk factors found in patient electronic health records (EHRs) and raw data from cardiac imaging studies, this software is able to more accurately predict VA risk in HCM patients than existing ACC, AHA, and ESC guidelines.
· Tested in a cohort of Johns Hopkins Medicine HCM patients, model accuracy improved when raw MRI data was analyzed in combination with EHR factors.
Stage of Development
· The technology is currently being tested by partner institutions for external validation.
Publication
· N/A
