Researchers at Wake Forest University School of Medicine have developed an AI model that identifies signs of heart failure from a standard electrocardiogram (ECG), including a form that often goes undetected in routine care. The model performed nearly as well using a single ECG lead - the same configuration captured by many smartwatches - suggesting it could eventually support more accessible screening.
Heart failure affects more than 6 million Americans and is a leading cause of hospitalization and death. Early detection usually requires an echocardiogram, a specialized imaging test that isn't available in every care setting. An AI-assisted ECG could help clinicians identify which patients need further evaluation.
What the AI model detects
The tool, described in the Journal of the American Heart Association, analyzes ECG data to flag three types of heart dysfunction: reduced ejection fraction (rEF), mildly reduced ejection fraction (mEF), and heart failure with preserved ejection fraction, or HFpEF. Ejection fraction measures the percentage of blood the heart's main pumping chamber pushes out with each beat. HFpEF is especially difficult to catch early and is often missed during routine clinical evaluations.
"This is a major step forward in how we can use everyday clinical tools to catch heart failure earlier," said Oguz Akbilgic, Ph.D., corresponding author and professor of artificial intelligence in the Department of Cardiovascular Medicine at Wake Forest University School of Medicine. "Our AI model can detect various types of heart dysfunction from a simple, single-lead ECG alone - the same lead configuration captured by many smartwatches and wearable ECG devices - suggesting the model could eventually be adapted for wearable-based screening."
How the study worked
Researchers trained the model on more than 1 million ECGs from Atrium Health Wake Forest Baptist, then tested it on a separate set of more than 72,000 ECGs from the University of Tennessee Health Science Center. The model classified ECGs into four categories: rEF, mEF, HFpEF, or no dysfunction. They tested two versions - one using 12-lead ECGs and one using a single lead, similar to what wearable devices collect.
The study adds to a growing body of AI for Science & Research work applying machine learning to clinical data.
Key findings
- Both models performed similarly. The 12-lead model was particularly effective at distinguishing patients with reduced ejection fraction from those without it. Its performance was lower, but still potentially useful, for the other two forms of heart dysfunction.
- The single-lead model performed nearly as well as the 12-lead model, suggesting the technology could eventually be adapted for wearable devices.
- In pediatric patients, the model detected reduced ejection fraction as well as or better than previously studied models, though the pediatric group was relatively small.
- The model generalized well across different demographic populations.
"Some of these conditions can progress without noticeable symptoms and may not be found until they become more severe," Akbilgic said. "Our model helps fill that gap by identifying electrical patterns in the heart that humans can't easily see so clinicians can decide when additional heart failure evaluation is needed."
What's next
The team is piloting the model in a family medicine clinic at Atrium Health Wake Forest Baptist to study how it performs when incorporated into clinical care.
"We're testing the tool in a real-world health care setting to determine whether it can help clinicians identify patients who need additional evaluation and how it might affect care and resource use," Akbilgic said.
Why this matters for healthcare and research professionals
For clinicians, the model is a concrete example of AI for Healthcare: it uses a test already routine in most practices to flag patients who need an echocardiogram before symptoms appear. For researchers, the single-lead result points toward wearable-based screening that could reach people who never see a cardiologist. The study was partially funded by the National Heart, Lung, and Blood Institute of the National Institutes of Health.
Your membership also unlocks:
