Jan. 20, 2026
Mayo Clinic Cardiovascular Medicine researchers developed a deep neural network using 12-lead electrocardiogram (ECG) results to quickly detect obstructive sleep apnea (OSA). The study was published in JACC: Advances.
The researchers used artificial intelligence (AI) algorithms to review the 12-lead ECG results of 11,299 patients who underwent sleep evaluations at Mayo Clinic.
The 12-lead ECGs are standard tests in medical practice. "Using the ECG to screen for OSA may be a novel, widely applicable low-cost strategy for identifying patients who may be helped by further evaluation, diagnosis and treatment of OSA. This study continues a series of advances from Cardiovascular Medicine and Mayo Clinic using AI to help improve patient care and further the science," says Virend Somers, M.D., Ph.D., a cardiologist and Alice Sheets Marriott Professor of Cardiovascular Medicine at Mayo Clinic in Rochester, Minnesota, and senior author of the study. Mayo Clinic continues to transform future possibilities in patient care with ongoing innovative solutions developed with Mayo's Artificial Intelligence (AI) in Cardiovascular Medicine specialty group.
Creating power tools
OSA is a common sleep-related breathing condition with cardiovascular consequences. Worldwide, more than 936 million adults ages 30 to 69 are affected by OSA. The often-missed diagnosis leaves a large proportion of patients with OSA undiagnosed, creating a need for high-performing screening tools.
This is why Mayo Clinic Cardiovascular Medicine researchers looked to develop a quick, effective and economical method for screening patients for OSA. "Because OSA can affect the heart, we hypothesized that we would be able to detect the presence of OSA using artificial intelligence algorithms to analyze the standard 12-lead ECG," says Dr. Somers. "Screening for OSA often requires questionnaires which are otherwise not routinely administered, and/or overnight testing with recordings of oxygen saturation and other measures. Having a simple measure like the ECG to identify those at high risk of OSA requires less time, labor and cost."
OSA's cardiovascular connection
Obstructive sleep apnea is a serious medical condition. Breathing stops and starts during sleep in patients with OSA. The throat muscles relax and block the airway. These pauses in breath, known as apneas, can occur many times during sleep. A common sign of OSA is snoring, though not everyone who snores has OSA.
The sudden drops in blood oxygen levels that occur during OSA increase blood pressure and strain the cardiovascular system. Patients with OSA may be at greater risk of:
- Coronary artery disease.
- Heart attack.
- Heart failure.
- Stroke.
- Arrhythmias.
- Sudden death.
A disease overlooked
Despite how common this serious medical condition is in adults, OSA is underdiagnosed and undertreated. Patients may not be aware of how they sleep or do not know that there may be reason for concern. "Patients aren't often asked about how they sleep unless they have a bed partner who proactively warns that they may have sleep apnea," says Dr. Somers.
The rate of OSA is rising in adults. "Obstructive sleep apnea in patients increases with obesity and with age. As obesity grows in prevalence, as it has until recently, and as the population ages, the occurrence of OSA increases," says Dr. Somers.
Obstructive sleep apnea does not present with the same symptoms in women and men. The disease is overlooked more often in women. "However, this ECG was more accurate in identifying OSA in women than in men, even though the women were younger and had less severe OSA. This suggests that the OSA cardiac 'fingerprint' is more evident in women than in men," says Dr. Somers.
Study highlights
The retrospective study population consisted of 11,299 patients ages 47 to 68 of whom 53.7% were male.
There were 7,170 patients with known OSA and 4,129 controls.
OSA was defined as an apnea-hypopnea index greater than or equal to 5.
Predictive performance of the algorithm in the total sample was evaluated using the receiver-operating characteristic curve with area under the curve (AUC). The AUC of the AI-ECG model for detecting OSA in the test sample was 0.80 with a 95% confidence interval (CI) of 0.77 to 0.83. Accuracy, sensitivity and specificity were 73.7%, 77.0%, and 68.6%, respectively. The diagnostic performance is similar or superior to that of the traditional OSA screening questionnaires or scores.
The model's predictive performance was also assessed separately in men and women. It showed better discriminatory performance in women (AUC: 0.82; 95% CI: 0.79-0.86) than in men (AUC: 0.73; 95% CI: 0.68-0.78).
Next steps
This research may lead to more tools in the field. "We were also able to pick up OSA using only 3 ECG leads rather than the standard 12-leads, so the screening may potentially be expanded to other ECG acquisition methods beyond the standard 12-lead ECG," says Dr. Somers.
In addition to offering an immediate and cost-effective screen for OSA, this AI-ECG model may help with decision-making and managing patient care. "We need to determine if the patients with a strong OSA signal on the ECG are more likely to develop significant cardiovascular disease in the future," says Dr. Somers. "This approach has the potential for helping identify those patients at risk of cardiac damage and possibly evaluate whether a given OSA treatment approach may be able to reduce their cardiovascular risk."
For more information
Covassin N, et al. Deep neural network algorithm using the electrocardiogram for detection of obstructive sleep apnea. JACC Advances. 2025;4:102139.
Refer a patient to Mayo Clinic.