Atrial fibrillation (AF) is a common heart condition—by 2030, the Centers for Disease Control estimates that nearly 12 million people will have AF. As one of the most common types of heart arrhythmia, AF occurs when the different chambers of the heart don’t beat in sync, causing a heartbeat that is either too fast, too slow, or just irregular overall. AF can increase the risk of clots and stroke, making it an important condition to detect and manage early.
According to a study published in Circulation, researchers at Massachusetts General Hospital, MIT, and Harvard have developed an artificial intelligence tool that could predict patients who might be at risk of developing AF. Early detection methods could allow for more timely and life-saving treatments and interventions.
Specifically, the AI tool built by the research team is designed to predict who is at risk of developing AF in the next 5 years. The training dataset the team provided for the AI tool to learn from included electrocardiogram test results from over 45,000 patients who received one at Massachusetts General.
Once trained, researchers then fed the AI tool additional datasets of electrocardiogram results from more than 83,000 patients to test its predictive capabilities. Study results found that the AI tool was able to effectively make predictions about patients who are at risk of developing AI, predictions with a similar use to standard models used by clinicians to detect AF risk. Co–lead author Shaan Khurshid, MD, MPH, highlighted how important these results are: “The application of such algorithms could prompt clinicians to modify important risk factors for atrial fibrillation that may reduce the risk of developing the disease altogether.”
The research team also heralded the use of artificial intelligence in medical practice and diagnostics, specifically, because it offers clinicians the ability to make medicine more precise. AI tools have been on the rise in medical practice for a range of different purposes, such as skin cancer detection. However, as researchers note, there are also ethical considerations to keep in mind due to the use of patient data.