A Study Of Deep Learning For Echo Analysis, Tracking, And Evaluation

DELINEATE-Deploy: Deep Learning for Echo Analysis, Tracking, and Evaluation - Prospective Deployment Platform IRB

The purpose of this study is to deploy and evaluate informational AI-Echo algorithms that assist echo clinicians in interpreting core echocardiographic parameters (e.g., LV/RV size and function, valvular disease severity) and stratifying disease progression risk. The primary outcome is clinician usability, interpretive consistency, and workflow integration. Second, we will conduct a pragmatic, stepped-wedge clinical trial with multiple arms evaluating diagnostic AI-Echo algorithms designed to identify specific cardiovascular diseases- such as genetic cardiomyopathy, ischemic heart disease, and cardiac amyloidosis-and assess whether AI deployment increases diagnostic testing and shortens time to diagnosis. Trials will be conducted using EHR-based notification systems with cluster-level randomization.

Inclusion criteria:

• Genetic cardiomyopathy arm: Clinicians who order, perform, and interpret echocardiograms and act on echocardiogram results, including both physicians and allied health staff taking care of patients aged ≥18 years who are undergoing a comprehensive TTE at a participating Mayo Clinic site with AI-Echo analysis indicating high risk for a genetic cardiomyopathy. A high-risk score will be defined by a specific threshold determined in model development to maximize sensitivity while maintaining an adequate positive predictive value to support clinical deployment
• Ischemic cardiomyopathy arm: Clinicians who order, perform, and interpret echocardiograms and act on echocardiogram results, including both physicians and allied health staff taking care of patients aged ≥18 years who are undergoing a comprehensive TTE with AI-Echo analysis indicating high risk for ischemic cardiomyopathy. A high-risk score will be defined by a specific threshold determined in model development to maximize sensitivity while maintaining an adequate positive predictive value to support clinical deployment.
• Cardiac amyloidosis arm: Clinicians who order, perform, and interpret echocardiograms and act on echocardiogram results, including both physicians and allied health staff taking care of patients aged ≥18 years who are undergoing a comprehensive TTE with AI-Echo analysis indicating high risk for cardiac amyloidosis. A high-risk score will be defined by a specific threshold determined in model development to maximize sensitivity while maintaining an adequate positive predictive value to support clinical deployment.
• Hypertrophic cardiomyopathy (HCM) arm: Clinicians who order, perform, and interpret echocardiograms and act on echocardiogram results, including both physicians and allied health staff caring for patients aged ≥18 years who are undergoing a comprehensive TTE at a participating Mayo Clinic site, with AI-Echo analysis indicating high risk for HCM. A high-risk score will be defined by a specific threshold determined during model development to maximize sensitivity while maintaining adequate positive predictive value for clinical deployment.

Exclusion criteria:

• Genetic cardiomyopathy arm: Studies performed within the past 2 years at a Mayo site or in those patients with known or suspected diagnosis of genetic cardiomyopathy under evaluation, on hospice care, or who have an expected non-cardiac life expectancy <1 year, and patients who have opted out of institutional and state research authorizations.
• Ischemic cardiomyopathy arm: Studies performed within the past 2 years at a Mayo site or in those patients with known CAD; prior myocardial infarction; revascularization with PCI or CABG; ischemic testing within the past 12 months; hospice care or expected non-cardiac life expectancy <1 year, and patients who have opted out of institutional and state research authorizations.
• Cardiac amyloidosis arm: Studies performed within the past 2 years at a Mayo site or in those patients with prior amyloid-specific testing (e.g., technetium pyrophosphate scan, cardiac MRI with late gadolinium enhancement suggestive of amyloid) or biopsy-proven systemic amyloidosis, on hospice care, or have expected non-cardiac life expectancy <1 year, and patients who have opted out of institutional and state research authorizations.
• Hypertrophic cardiomyopathy (HCM) arm: Studies performed within the past 2 years at a Mayo site or patients with a known diagnosis of HCM documented in the medical record prior to the index TTE, prior septal reduction therapy (surgical myectomy or alcohol septal ablation), or patients on hospice care or with an expected non-cardiac life expectancy <1 year, and patients who have opted out of institutional and state research authorizations.