FDA Clears EchoNext AI for Structural Heart Disease Detection on Routine ECGs

EchoNext was developed at Columbia and NewYork-Presbyterian by Pierre Elias and cleared in June 2025 for right- and left-sided heart failure, valvular disease, hypertrophic cardiomyopathy, and pulmonary hypertension. The model converts the ubiquitous 12-lead ECG into a pretest probability tool that triggers echocardiography, addressing the absence of any population screening test for the leading global cause of death. Retrospective training data showed it identified severe undiagnosed cardiomyopathy that later required transplant in at least one documented case published in Nature Medicine.

The Healthline account omitted that performance was measured on retrospective rather than prospective data and that the 13-percentage-point gain over cardiologists was achieved without external validation cohorts. No randomized trial has yet tested whether earlier echo referral changes mortality or hospitalization rates. Funding from Pathway Labs and the absence of an independent test set raise the usual concerns about spectrum bias and overfitting seen in other ECG-AI products.

Next steps include ongoing multi-center prospective studies and integration into existing ECG carts. If those trials meet pre-specified positive-predictive-value thresholds above 0.40 in primary-care populations, health systems could begin routing flagged patients to echocardiography within routine visits by late 2026.

Regulatory clearance rests on substantial equivalence to predicate software rather than new outcome data, so real-world impact remains to be measured.