[ad_1]
January 08, 2026
2 min read
Key takeaways:
- An AI model accurately detected features of myocardial ischemia and coronary microvascular and vasomotor dysfunction on ECGs.
- Diagnosis previously required specialty-specific imaging techniques.
Researchers developed an AI model to detect myocardial ischemia and coronary microvascular and vasomotor dysfunction using data from an ECG strip; the diagnosis previously required advanced imaging techniques.
Although additional clinical trials are needed, models such as this may improve chest pain diagnosis in resource-limited or nonspecialty centers, the researchers wrote.
An AI model accurately detected features of myocardial ischemia and coronary microvascular and vasomotor dysfunction on ECGs. Image: Adobe Stock
Findings from initial development and validation of an AI model to detect myocardial ischemia and coronary microvascular and vasomotor dysfunction were published in NEJM AI.
“Our model creates a way for clinicians to accurately identify a condition that is notoriously hard to diagnose — and often missed in emergency department visits — using a 10-second [ECG] strip,” Venkatesh L. Murthy, MD, PhD, Melvyn Rubenfire Professor of Preventive Cardiology at University of Michigan Medical School and associate chief of cardiology for translational research and innovation at University of Michigan Health Frankel Cardiovascular Center, said in a press release. “Essentially, we taught the model to ‘understand’ the electrical language of the heart without human supervision.”
Researchers developed and pretrained a self-supervised learning model using a database of 800,035 unlabeled ECG waveforms, and fine-tuned it using smaller, labeled PET scan databases for 12 demographic and clinical prediction tasks across three dimensions: myocardial function, coronary perfusion and cardiac rhythm.
According to the study, four of the prediction tasks focused on assessing for myocardial ischemia and coronary microvascular and vasomotor dysfunction, which were inaccessible using prior ECG models and typically required advanced modalities such as PET myocardial perfusion imaging.
Model accuracy and generalizability were evaluated across five additional databases, including the publicly available PTB-XL and UK Biobank databases, according to the study.
Model performance varied across diagnostic tasks, with area under the receiver operator curve (AUROC) ranging from 0.763 for the detection of impaired myocardial flow reserve to 0.955 for impaired left ventricular ejection fraction (defined as < 35%), according to the study.
The researchers reported that the model’s performance remained strong during validation in three external and two internal bases and improved further with self-supervised learning, with the AUROC ranging from 0.771 for impaired myocardial flow reserve to 0.949 for impaired LVEF.
“In hospitals with limited resources or nonspecialty centers, using our [ECG]-AI model to predict myocardial flow reserve and [coronary microvascular dysfunction] will be an easy, cost-effective and noninvasive way to identify when a patient would benefit from advanced testing for a serious condition,” Sascha N. Goonewardena, MD, associate professor of internal medicine-cardiology at University of Missouri Medical School, said in the release.
Sources/Disclosures
Source:
Moody JB, et al. NEJM AI. 2025;doi:10.1056/AIoa2500164.
Reference:
Disclosures:
Goonewardena reports no relevant financial disclosures. Murthy reports owning stock in Abbott Laboratories, AbbVie, Advanced Micro Devices, Amgen, Baxter International, Boston Scientific, Bristol Myers Squibb, Cardinal Health, Cigna, Eli Lilly, General Electric, Ionetix, Johnson & Johnson, Medtronic, Merck, Nvidia, Organon, Pfizer, Viatris, Zimmer Biomet Holdings and ZimVie; and consulting for INVIA Medical Imaging Solutions, Ionetix, Jubilant DraxImage Inc. and Siemens Medical Solutions USA. Please see the study for all other authors’ relevant financial disclosures.
Ask a clinical question and tap into Healio AI’s knowledge base.
- PubMed, enrolling/recruiting trials, guidelines
- Clinical Guidance, Healio CME, FDA news
- Healio’s exclusive daily news coverage of clinical data
[ad_2]
Source link
