May 7 2021
There are several forms of heart disease, but some types of this medical condition, like asymptomatic low ejection fraction, can be difficult to detect, particularly during the early stages when therapies would be highly effective.
The EAGLE—short for ECG AI-Guided Screening for Low Ejection Fraction—trial attempted to establish whether an artificial intelligence (AI) screening tool designed to spot low ejection fraction through EKG data could enhance the diagnosis of this medical disorder in regular practice. The results of the study have been published in the Nature Medicine journal.
Systolic low ejection fraction refers to the inability of the heart to contract sufficiently with each heartbeat to pump a minimum of 50% of the blood from its chamber. While an echocardiogram can instantly diagnose low ejection fraction, this time-intensive imaging test needs more resources when compared to a 12-lead EKG, which is cost-effective, fast, and instantly available. The AI-based EKG algorithm was tested and designed via a convolutional neural network and verified in subsequent analyses.
The EAGLE trial was conducted in 45 medical institutions based in Wisconsin and Minnesota, including academic medical centers, community medical centers, and rural clinics. On the whole, 348 primary care clinicians from 120 medical care teams were arbitrarily assigned to general intervention or care.
The intervention team was alerted to a positive screening outcome for low ejection fraction through the electronic health record and this prompted them to order an echocardiogram to validate.
The AI-enabled EKG facilitated the diagnosis of patients with low ejection fraction in a real-world setting by identifying people who previously would have slipped through the cracks.
Peter Noseworthy, MD, Study Senior Author and Cardiac Electrophysiologist, Mayo Clinic
During the trial, 22,641 adult patients had an EKG under the supervision of the clinicians for a period of eight months. The AI detected positive results in 6% of the patients. Overall, the number of patients who received an echocardiogram was quite similar, but among patients who had a positive screening result, echocardiogram was performed on a greater percentage of the intervention group.
The AI intervention increased the diagnosis of low ejection fraction overall by 32% relative to usual care. Among patients with a positive AI result, the relative increase of diagnosis was 43%. To put it in absolute terms, for every 1,000 patients screened, the AI screening yielded five new diagnoses of low ejection fraction over usual care.
Xiaoxi Yao, PhD, Study First Author and Health Outcomes Researcher in Cardiovascular Diseases, Mayo Clinic
“With EAGLE, the information was readily available in the electronic health record, and care teams could see the results and decide how to use that information. The takeaway is that we are likely to see more AI use in the practice of medicine as time goes on. It’s up to us to figure how to use this in a way that improves care and health outcomes but does not overburden front-line clinicians,” added Dr Noseworthy.
The EAGLE trial also employed a positive deviance method to assess the top five users and the top five non-users of the AI screening data among the care team. According to Dr Yao, this cycle of feedback and learning from doctors will reveal new ways to enhance the application and adaptation of the AI tool in practice.
EAGLE is one of the first large-scale trials to show the usefulness the AI technology in regular practice. The low ejection fraction algorithm has received breakthrough designation from Food and Drug Administration and is one of the many algorithms designed by Mayo Clinic and licensed to Anumana Inc.—a new firm focused on revealing the mysterious biomedical knowledge to facilitate early detection and speed up treatments for heart disease.
Moreover, the low ejection fraction algorithm was also formerly licensed to Eko Devices Inc., particularly for portable devices that are externally used on the chest.
The EAGLE trial was financially supported by Mayo Clinic’s Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery, in association with the departments of Cardiovascular Medicine and Family Medicine, and the Division of Community Internal Medicine.
Yao, X., et al. (2021) Artificial intelligence-enabled electrocardiograms for identification of patients with low ejection fraction: a pragmatic, randomized clinical trial. Nature Medicine. doi.org/10.1038/s41591-021-01335-4.