Two new artificial intelligence products that can improve cardiac care
Two new artificial intelligence products that can improve cardiac care
https://employercoverage.substack.com/p/two-new-artificial-intelligence-products
Publish Date: 2026-07-01 05:11:00
Source Domain: employercoverage.substack.com
Summary: Two AI tools, one that detects structural heart disease from EKGs and one that identifies which heart failure patients need implantable defibrillators, show promise for improving cardiac care, though employers should expect these to raise costs.
Two recent studies illustrate how AI can sharpen cardiac diagnosis and guide treatment decisions for patients with, or at risk for, serious heart disease
The first is a report that EchoNext, an AI algorithm that analyzes electrocardiograms (EKGs), will become available to clinicians through OpenEvidence, a clinical decision-support platform built on large language models and trained on high-impact medical literature. EchoNext, which was developed at New York Presbyterian Hospital and Columbia University, screens EKGs for subtle signs of structural heart disease, such as valve problems or heart failure. A 2025 publication in Nature Medicine found that EchoNext outperformed cardiologists in identifying structural heart disease across a standardized EKG dataset. This is an example of using AI to increase the yield from studies that are already being performed — similar to recent reports of using AI to detect cardiovascular disease on mammograms. Broader access through OpenEvidence could extend this capability to clinicians who lack routine access to specialized cardiology interpretation.
The second is a report from Nature that researchers have used deep learning to identify a “biomarker” for patients that would benefit from implantable defibrillators. People with heart failure and structural heart disease are at elevated risk of life-threatening arrhythmias. An implantable cardioverter-defibrillator (ICD) can deliver an electric shock to restore normal rhythm and prevent sudden cardiac death. Currently, patients with low ejection fractions — the share of blood the left ventricle pumps with each heartbeat — are typically recommended for an ICD, but this new deep-learning algorithm proved more accurate than…
