2024-09-03 15:10:48
(Montreal) A recent study by researchers at the Montreal Heart Institute shows that an artificial intelligence algorithm can very accurately detect future atrial fibrillation from a simple electrocardiogram.
Published at 11:10 a.m., updated at 1:32 p.m.
Jean-Benoit Legault Canadian Press
Atrial fibrillation (AF) is the most common form of sustained cardiac arrhythmia but is usually asymptomatic. If left untreated, it significantly increases the risk of stroke, cognitive decline and dementia.
“There is so much information hidden in an ECG that even an expert’s eye cannot tell whether a patient is developing an arrhythmia,” explains Dr. Rafik Tadros, a cardiologist, arrhythmia specialist and Canada Research Chair. PhD.
“So the intelligence applied to the ECG actually allows us to extract all the data from the ECG and predict the future occurrence of atrial fibrillation very accurately and dramatically. »
Therefore, the use of this algorithm will represent a major advance in the early detection of atrial fibrillation and the prevention of its complications, the press release explains.
Dr. Tadros and his colleagues used nearly a million electrocardiograms from some 100,000 patients at the Heart Institute to “teach” an artificial intelligence tool to detect sudden atrial fibrillation. The tool’s accuracy was subsequently confirmed in a second cohort at a US hospital.
“This shows us that this model does work, not just in the ICM population but in other populations as well,” Dr. Tadros said.
The study also showed that deep learning provides superior performance in predicting atrial fibrillation compared to traditional clinical prediction and genetic prediction.
Dr. Tadros and his collaborators are now discussing clinical deployment of their tool, which could happen soon. Once the tool is deployed, the electrocardiograms of all patients presenting to the ICM, regardless of their health problems, will be analyzed for signs of current or future atrial fibrillation.
“For example, for breast or colon cancer screening, it’s the same principle,” explains Dr. Tadros. Initially it is asymptomatic and if we catch it early we can prevent complications. »
He added that patients identified by the AI tool can have their electrocardiograms closely monitored for two weeks using a device they wear at all times to detect intermittent atrial fibrillation.
More than 2% of people aged 65 years have atrial fibrillation, rising to 10% of people aged 80 years and older.
The research was led by graduate student Gilbert Jabbour, Ph.D., under the guidance of Robert Avram, Ph.D., artificial intelligence researcher and director of the HeartWise.ai Lab, and Tadros, Ph.D. All are at the Montreal Heart Institute.
The research results are provided by European Heart Journal. They are based on the analysis of more than 1 million electrocardiograms from more than 250,000 patients. The findings were also presented at the European Congress of Cardiology in London on September 1.
1725388400
#Montreal #Heart #Institute #study #Artificial #intelligence #detect #atrial #fibrillation
