Atrial fibrillation — an uneven and often rapid heart rate — is a frequently occurring condition that could, at times, lead to the development of clots in the heart that can then travel to the brain, resulting in a stroke.
As illustrated in a study published in Circulation, a team directed by scientists at Massachusetts General Hospital (MGH), the Broad Institute of MIT and Harvard has formulated an artificial intelligence (AI)-based technique for detecting patients who are at risk for developing atrial fibrillation and could thus gain from preventative measures.
The team formulated the AI-based technique to estimate the risk of atrial fibrillation in the span of the next five years based on the outcomes from electrocardiograms (noninvasive tests that record the heart’s electrical signals) in 45,770 patients undergoing primary care at MGH.
Next, the researchers applied their technique to three large data sets from studies comprising a total of 83,162 individuals. The AI-based technique estimated atrial fibrillation risk on its own and was synergistic when integrated with established clinical risk factors for estimating atrial fibrillation. The technique was also exceptionally predictive in subsets of individuals, such as those who had a prior stroke or heart failure.
We see a role for electrocardiogram-based artificial intelligence algorithms to assist with the identification of individuals at greatest risk for atrial fibrillation.
Steven A. Lubitz, MD, MPH, Senior Author and Cardiac Electrophysiologist, MGH
Steven A. Lubitz is also an associate member at the Broad Institute.
Co-lead author Shaan Khurshid, MD, MPH, an electrophysiology clinical and research fellow at MGH states: “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.”
Lubitz explains that the algorithm could act as a form of pre-screening tool for patients who may presently be undergoing unnoticed atrial fibrillation, stimulating clinicians to look for atrial fibrillation using longer-term cardiac rhythm monitors, which could, in turn, result in stroke prevention strategies.
The findings of this study also show the potential power of AI — which in this case involves a particular type called machine learning — to progress medicine.
With the explosion of data science technologies and the vast amounts of clinical data now available, machine learning is poised to help clinicians and researchers make great strides in enhancing cardiology care.
Anthony Philippakis, MD, PhD, Study Co-Author and Chief Data Officer, Broad Institute
Anthony Philippakis is also the co-director of the institute’s Eric and Wendy Schmidt Center.
“As a data scientist and former cardiologist, I’m excited to see how machine learning-based methods can work with the tests and clinical approaches we use every day to help us improve risk prediction and take care of patients with atrial fibrillation,” Anthony Philippakis added.
Lubitz is an associate professor of Medicine at Harvard Medical School.
The other co-authors are Samuel Friedman, PhD, Christopher Reeder, PhD, Paolo Di Achille, PhD, Nathaniel Diamant, BS, Pulkit Singh, BS, Lia X. Harrington, PhD, Xin Wang, MBBS, MPH, Mostafa A. Al-Alusi, MD, Gopal Sarma, MD, PhD, Andrea S. Foulkes, ScD, Patrick T. Ellinor, MD, PhD, Christopher D. Anderson, MD, MMSc, Jennifer E. Ho, MD and Puneet Batra, PhD.
This study received support from the National Institutes of Health, the Doris Duke Foundation, the American Heart Association and the Leducq Foundation.
Khurshid, S., et al. (2021) Electrocardiogram-based Deep Learning and Clinical Risk Factors to Predict Atrial Fibrillation. Circulation. doi.org/ 10.1161/CIRCULATIONAHA.121.057480.