The electrocardiogram (ECG) has been a standard procedure in medicine for more than a hundred years – inexpensive, quick and available everywhere. With artificial intelligence it is experiencing a renaissance: algorithms recognise patterns in the tracings that remain hidden even to experienced cardiologists.
Hidden heart failure in the ECG
In heart failure with impaired pumping function (reduced ejection fraction), the heart can no longer pump enough blood around the body. It is often diagnosed late because early symptoms are non-specific. Until now, an echocardiogram (heart ultrasound) was usually needed for the diagnosis. Researchers at the Mayo Clinic trained an AI to recognise weak pumping function from a standard 12-lead ECG.
In the randomised EAGLE trial (published in Nature Medicine in 2021), the system was tested in routine primary care: 348 doctors in 120 practice teams cared for 22,641 adults who had an ECG. In the group with an AI alert, 32 per cent more new cases of heart failure with reduced ejection fraction were diagnosed than in the comparison group. In absolute terms, there were about five additional diagnoses per 1,000 people examined.
Atrial fibrillation via smartwatch
Atrial fibrillation is the most common sustained heart rhythm disorder and significantly increases the risk of stroke. Because it often causes no symptoms, it frequently goes undetected. Many smartwatches and fitness bands analyse the pulse using algorithms and can record a single-lead ECG.
In the Apple Heart Study with 419,297 participants (published in the New England Journal of Medicine in 2019), 0.52 per cent received a notification of an irregular pulse. When a notification occurred at the same time as monitoring with an ECG patch during follow-up, atrial fibrillation was confirmed in 84 per cent of cases.
Important: whether smartwatch screening prevents strokes has not been conclusively demonstrated. An alert from the watch does not replace a diagnosis, and a normal reading does not rule out heart problems.
Further areas of application
- Echocardiography: AI helps with image acquisition and automatically measures chamber sizes and pumping function.
- Imaging: in CT angiography of the coronary arteries, algorithms support the assessment of narrowings and plaques.
- Risk prediction: models attempt to estimate the risk of heart attack or heart failure from the ECG, laboratory values and medical history more accurately than conventional risk calculators.
What wearable users should know
- If you receive an alert about atrial fibrillation, the recording should be assessed by a doctor – no need to panic, but promptly.
- Chest pain, shortness of breath, fainting or sudden paralysis are always an emergency (112 in Europe), regardless of what the watch shows.
- Frequent false alarms can be unsettling. If you notice that constant self-monitoring is stressful, you can switch off notifications and discuss this with your doctor.
Conclusion
Cardiology shows how AI can get more out of simple, established examinations. The EAGLE trial is an example of how benefit in practice should be tested. For widespread use – for example as screening of the whole population – evidence is still lacking that this actually prevents the consequences of disease.
More in our encyclopaedia: Heart failure and Atrial fibrillation.
Note: This article is for general information only and does not replace medical advice. As of: October 2026.
Video on this article
Watch on YouTube · Subscribe to the channel
Sources
- Mayo Clinic: Trial demonstrates early AI-guided detection of heart disease in routine practice (EAGLE, Nature Medicine 2021)
- Yao X et al.: Artificial intelligence–enabled electrocardiograms for identification of patients with low ejection fraction: a pragmatic, randomized clinical trial. Nature Medicine 2021
- Perez MV et al.: Large-Scale Assessment of a Smartwatch to Identify Atrial Fibrillation. New England Journal of Medicine 2019
- Deutsche Gesellschaft für Kardiologie / ESC: Leitlinien zu Vorhofflimmern und Herzinsuffizienz
Leave a Reply