Level 1 - Absolute Beginner
Doctors use a heart test called an ECG. It checks the heart's electrical signals. Now a new computer tool can read this test very well.
Scientists at Wake Forest University made the tool. It uses artificial intelligence, or AI. AI is a computer program that learns from a lot of data.
The tool looks for heart failure. Heart failure means the heart does not pump blood well. One type of heart failure is very hard to find. Doctors often miss it.
The tool learned from more than 1 million heart tests. Then it was tested on 72,000 new tests from a different hospital. It worked well. It may help doctors find heart problems early.
- artificial intelligence (AI)
- A computer program that can learn and find patterns.
- electrocardiogram (ECG)
- A test that checks the heart's electrical signals.
- ejection fraction
- A number that shows how well the heart pumps blood.
- heart failure
- When the heart cannot pump blood well.
- HFpEF
- A type of heart failure that is hard for doctors to find.
- echocardiogram
- A test that uses sound waves to make pictures of the heart.
- wearable device
- A small device you wear, like a smartwatch.
- screening
- A quick test to check for a health problem early.
Level 2 - Elementary
A new artificial intelligence (AI) tool can read a common heart test called an electrocardiogram, or ECG. An ECG records the heart's electrical activity using small sensors placed on the skin.
Researchers at Wake Forest University School of Medicine built the tool. It looks for three kinds of heart dysfunction, including a type called HFpEF, which is often missed during normal checkups.
The team trained the AI using more than 1 million ECGs from one hospital system. Then they tested it on over 72,000 ECGs from a different hospital, to make sure it worked well on new patients it had never seen before.
The tool even performed well using data from just one ECG lead, similar to what some smartwatches can record. This could make cheap, easy heart screening possible in more places, including small clinics.
- artificial intelligence (AI)
- Computer software that learns from large amounts of data to make predictions.
- electrocardiogram (ECG)
- A test that records the heart's electrical activity using sensors placed on the skin.
- ejection fraction
- A measurement of how much blood the heart pumps out with each beat.
- heart failure
- A condition in which the heart cannot pump enough blood for the body's needs.
- HFpEF
- Heart failure with preserved ejection fraction, a type that is often missed because the heart still pumps a normal amount of blood.
- echocardiogram
- An ultrasound test that creates moving images of the heart.
- wearable device
- A gadget like a smartwatch that can track health information on the body.
- screening
- Testing people who feel healthy to catch a disease before symptoms appear.
Level 3 - Intermediate
Researchers at Wake Forest University School of Medicine have developed an artificial intelligence model capable of interpreting a routine electrocardiogram (ECG) to identify several forms of heart dysfunction, including one, heart failure with preserved ejection fraction (HFpEF), that frequently escapes detection in standard clinical practice.
Published in August 2026 in the Journal of the American Heart Association, the study describes how the model was trained on more than 1 million ECGs from Atrium Health Wake Forest Baptist before being validated on an independent set of over 72,000 ECGs collected at the University of Tennessee Health Science Center, a step designed to confirm that its accuracy would generalize beyond the population it was trained on.
Perhaps the most striking finding is that the model performed well even when given data from a single ECG lead, comparable to the readings captured by some consumer wearable devices, raising the prospect of inexpensive screening tools that do not require a full twelve lead ECG machine.
Because a definitive diagnosis of heart dysfunction typically requires an echocardiogram, a specialized ultrasound exam that is not universally available, clinicians hope that an ECG based AI screen could extend early detection into smaller clinics and even onto people's wrists, at a stage when treatment options are more likely to be effective.
- artificial intelligence (AI)
- Computer systems trained on large datasets to recognize patterns and make predictions without being explicitly programmed for each task.
- electrocardiogram (ECG)
- A diagnostic test that records the heart's electrical activity through electrodes placed on the skin.
- ejection fraction
- A percentage that indicates how much blood the left ventricle pumps out with each contraction.
- heart failure
- A chronic condition in which the heart's pumping ability is impaired, reducing the body's supply of oxygen rich blood.
- HFpEF
- Heart failure with preserved ejection fraction, a subtype in which the heart still pumps a near normal percentage of blood but cannot relax and fill properly, making it notoriously difficult to diagnose.
- echocardiogram
- An ultrasound based imaging test used to directly visualize the heart's structure and function.
- wearable device
- A body worn electronic device, such as a smartwatch, capable of recording basic physiological data.
- screening
- The practice of testing asymptomatic individuals to detect disease at an early, more treatable stage.
Level 4 - Advanced
A team of researchers at Wake Forest University School of Medicine has trained an artificial intelligence model to extract, from an ordinary electrocardiogram, signals of cardiac dysfunction that routinely elude clinicians, most notably heart failure with preserved ejection fraction (HFpEF), a form of the disease whose subtlety on conventional testing has long made it a diagnostic blind spot.
The findings, published in the Journal of the American Heart Association in August 2026 under the title ECG based artificial intelligence for classifying left ventricular dysfunction and heart failure, describe a model trained on more than 1 million ECGs drawn from Atrium Health Wake Forest Baptist and subsequently validated against an entirely separate cohort of over 72,000 ECGs from the University of Tennessee Health Science Center, a design intended to test whether the model's performance would hold up outside the institution and population that produced it.
What distinguishes the result is not merely accuracy but accessibility: the model retained its performance when fed data from a single ECG lead, the sort of stripped down signal already captured by some consumer wearables, hinting at a future in which meaningful cardiac screening might not require specialized equipment at all.
That matters because a conclusive diagnosis of heart dysfunction still generally depends on an echocardiogram, an ultrasound based imaging test unavailable in many clinics, so a reliable ECG derived screen could push early detection, and with it the better outcomes that early treatment makes possible, into settings and populations that specialist imaging currently does not reach.
- artificial intelligence (AI)
- Computational models trained on vast datasets to detect patterns and generate predictions that would be difficult or impossible for humans to derive manually.
- electrocardiogram (ECG)
- A standard diagnostic test that records the heart's electrical activity via electrodes on the skin, traditionally interpreted for rhythm and rate rather than pumping function.
- ejection fraction
- A percentage measure of the volume of blood the heart's left ventricle expels with each contraction, long used as a clinical proxy for cardiac function.
- heart failure
- A progressive syndrome in which the heart's pumping capacity deteriorates, depriving tissues of adequate blood flow and oxygen.
- HFpEF
- Heart failure with preserved ejection fraction, a form in which the heart's ejection percentage appears normal even as its ability to relax and fill with blood is impaired, a discrepancy that has made it exceptionally prone to underdiagnosis.
- echocardiogram
- An ultrasound based imaging study considered the diagnostic standard for directly visualizing cardiac structure and function, though not universally accessible.
Talk about this article with KatePractice this story's vocabulary in a real spoken conversation with your AI tutor