Early research suggests the use of AI to predict heart attack and stroke risk using a single chest x-ray shows promise

CNN
—

Early research suggests a promising way to use artificial intelligence to predict 10-year risk of dying from a heart attack or stroke from a single chest x-ray.

of Preliminary findings It was announced Tuesday at the annual meeting of the Radiological Society of North America. The study is in the final draft stage and has not yet been submitted for publication in a medical journal.

Researchers used approximately 150,000 chest x-rays to train an artificial intelligence program to identify patterns in the images associated with risk from major cardiovascular disease events. They tested the program on a separate group of about 11,000 people and found a “significant association” between the risk levels predicted by the AI ​​and the actual occurrence of major cardiovascular disease events.

The clinical criterion for analyzing cardiovascular disease risk is the atherosclerotic cardiovascular disease (ASCVD) risk score. This is a calculator that weights various patient data points known to be highly associated with adverse cardiovascular events, such as age, blood pressure, and history of cardiovascular disease. smoking.

Statins are recommended for people with a 10-year risk of 7.5% or higher. The AI ​​model uses the same risk thresholds as established risk calculators, and early findings suggest it performs similarly.

“Although it has long been recognized that X-rays capture information beyond traditional diagnostic findings, this data has not been used due to the lack of robust and reliable methods. Massachusetts General Hospital and Harvard Participating in the AI ​​in Medicine program at Brigham and Women’s Hospital, a medical school.

AI findings may match traditional radiology measurements, but they may also detect things that may have gone unnoticed.

“Some of them are anatomical changes that we can see with the naked eye, and they are physiologically meaningful. However, I believe that much of the information captured or extracted is embedded in the scan, which, as a traditionally trained radiologist, I cannot, for now, make sense of.

“It has this black-box character,” he said.

Donald Lloyd Jones, Ph.D., chair of preventive medicine at Northwestern University Feinberg School of Medicine and past president of the American Heart Association, co-chaired the Risk Assessment Committee when the ASCVD risk calculator was created in 2013. 2018, when the guidelines were updated to emphasize the relationship between risk scores and personal medical history.

He isn’t involved in new AI research, but says it’s important to keep the field moving forward.

“This is exactly the application that artificial intelligence is best suited for,” he said. “So we need to keep doing things like this, especially to really understand if we can find patients who would otherwise be overlooked.

However, it is still important to collect all patient data points that are input into established risk calculators. Because they are practical. And whether risk is calculated using statistical formulas or AI models, individualized patient assessment is required for the most effective results.

“A chest x-ray will not cure smoking. Some, but not all, it’s a process in which both the patient and the doctor discuss the patient’s risks and how well they think the statin will help them.”

For the study, Weiss and co-authors trained an AI using chest x-rays from participants in the National Cancer Institute’s prostate, lung, colorectal, and ovarian cancer screening trials. Did. It was tested at Mass General Brigham in people with a mean age of 60 who had routine outpatient chest x-rays and who were potential candidates for statin therapy.

Additional studies involving controlled randomized trials are needed to validate deep learning models.