Atrial fibrillation (AFib) is an irregular heartbeat or arrhythmia, the most common type of heart rhythm disorder and while it usually doesn’t have harmful consequences by itself, the real danger is an increased risk for stroke. About one in three people with AFib aren’t even aware that they have the condition.
Researchers at Cedars-Sinai Medical Center’s Smidt Heart Institute in Los Angeles are beginning to tackle this concern with an algorithm they developed to detect abnormal heart rhythm in symptomless people. Cedars-Sinai published their research study last month in JAMA Cardiology and found that their algorithm can successfully identify signals from electrocardiograms (ECGs) that demonstrate a risk of AFib.
Neal Yuan, M.D., director of cardiac rehabilitation at the San Francisco VA Health System and one of the researchers of the study, explained that the algorithm makes predictions using “a complicated equation involving the 20,000 values that make up an ECG.” Data from each ECG are entered into the deep learning model to test the prediction ability of the equation and make modifications as needed based on the validity of the predictions. The algorithm’s equation was calibrated with nearly 1 million routine ECGs from six different Veteran Affairs hospitals and Cedars-Sinai Medical Center.
To ensure accurate performance across populations, Cedars-Sinai tested the model on several diverse patient populations, including women; Black, brown and other people of color; patients younger than 65 and those older than 65; patients with few medical comorbidities as well as patients with multiple comorbidities. Yuan said that it “performed just as well in these different subgroups and across different medical sites, which reassured us that our model was picking up on features in the ECG that are widely generalizable across all patients.”
According to Yuan, more studies are in the works to deploy artificial intelligence models, such as Cedar-Sinai’s, into clinical settings to assess how these algorithms will affect physician practices and patient outcomes. With this new tool, heart disorders can be predicted earlier and with more accuracy to ensure better outcomes for those with AFib and other heart conditions.
