For the first time, researchers at Yale University were able to diagnose structural heart disease using a simple electrocardiogram (ECG) sensor in a smartwatch. Their findings will be presented this week at the American Heart Association’s 2025 Scientific Sessions.
Wearable devices are increasingly being used to diagnose and monitor heart conditions that affect heart rhythm, such as arrhythmias and atrial fibrillation, but single-lead ECG designs have so far been limited in their ability to diagnose physical defects in the heart. However, with the help of artificial intelligence (AI) algorithms, a group of 600 patients in a real hospital environment were able to be diagnosed with high accuracy with just 30 seconds of smartwatch reading.
“Single-lead ECGs by themselves are limited; they cannot replace the 12-lead ECG tests available in a healthcare setting. However, using AI makes them powerful enough to screen for important heart conditions,” said Rohan Khera, MD, senior author of the study and director of the Cardiovascular Data Science Institute at Yale School of Medicine. “This will allow us to perform early screening for structural heart disease at scale using a device that many people already own.”
The AI algorithm was developed using more than 266,000 complete ECGs from 110,000 patients at Yale New Haven Hospital from 2015 to 2023. Khera and colleagues isolated one of 12 leads that most closely resembled a single lead found in a smartwatch sensor and used that data to predict the presence of three types of structural heart disease: low left ventricular ejection fraction, severe left valvular disease, and severe left ventricular disease. Hypertrophy.
The model was first validated in more than 44,000 adults from four regional hospitals and 3,000 participants in a population-based study in Brazil. A prospective study then recruited 600 patients undergoing echocardiography and took 30-second ECG measurements using a smartwatch before the procedure.
In this patient cohort, the algorithm achieved 86% sensitivity and 87% specificity, with overall performance of 88% using single-lead readings and 92% using 12-lead hospital ECG equipment. This was achieved by introducing noise into the data used to train the model, which helped increase the reliability of the AI when analyzing ECG signals obtained in real-world situations.
“We investigated whether the same smartwatches people wear every day could help detect hidden structural heart disease early before it progresses to serious complications or heart disease,” said Arya Aminorroaya, MD, an internal medicine resident at Yale New Haven Hospital and a researcher at Yale School of Medicine. “Millions of people wear smartwatches, and they are now primarily used to detect heart rhythm problems such as atrial fibrillation. Structural heart disease, on the other hand, is typically discovered with echocardiography, an advanced echocardiographic imaging test that requires special equipment and is not widely available for routine screening.”
Although promising, these results need to be validated in a larger population. In this prospective study, only 5% of patients had structural heart disease confirmed during ultrasound procedures. Fifteen patients had low left ventricular ejection fraction, five were diagnosed with severe left valvular disease, and only one patient had severe left ventricular hypertrophy.
“We plan to explore ways in which AI tools can be evaluated in a broader setting and integrated into community-based cardiology screening programs to assess their potential impact on improving preventive care,” Aminorroaya said.
