Artificial Intelligence Can Use Routine ECGs to ID Heart Disease
FRIDAY, Jan. 11, 2019 — Artificial intelligence (AI) can identify asymptomatic left ventricular dysfunction (ALVD) using results from a routine electrocardiogram (ECG), according to a research letter published online Jan. 7 in Nature Medicine.
Zachi I. Attia, from the Mayo Clinic in Rochester, Minnesota, and colleagues trained a convolutional neural network to identify ALVD using routine ECGs from 44,959 patients. The model was then tested in an independent set of 52,870 patients. Ventricular dysfunction was defined as ejection fraction ≤35 percent.
The researchers found that the network model yielded an area under the curve of 0.93 as well as sensitivity, specificity, and accuracy of 86.3, 85.7, and 85.7 percent, respectively. In patients without ventricular dysfunction, the risk for developing future ventricular dysfunction was four times higher in those with a positive AI screen compared with those with a negative screen (hazard ratio, 4.1).
“The ability to acquire an ubiquitous, easily accessible, inexpensive recording in 10 seconds — the EKG — and to digitally process it with AI to extract new information about previously hidden heart disease holds great promise for saving lives and improving health,” a coauthor said in a statement.
Abstract/Full Text (subscription or payment may be required)
Posted: January 2019
Source: Read Full Article