AI algorithm accurately predicts risk of heart attack within 5 years

 

The AI tool can do in seconds what previously took experts up to 30 minutes

Depositphotos

A new AI tool developed by researchers from Cedars-Sinai Medical Center can accurately measure plaque deposits in coronary arteries and predict a patient’s risk of suffering a heart attack within five years. The tool needs further validation before being deployed in clinics, but promises to automatically achieve in seconds what has previously taken trained experts up to 30 minutes to deliver.

A scan known as computed tomography angiography (CTA) is one of the best tools doctors currently have at their disposal to evaluate heart disease patients. CTA imaging of plaque deposits within coronary arteries has recently been found to be the best way to predict a patient’s likelihood of heart attack in the near future.

"Coronary plaque is often not measured because there is not a fully automated way to do it," said senior author of the new study, Damini Dey, from the Biomedical Imaging Research Institute at Cedars-Sinai. "When it is measured, it takes an expert at least 25 to 30 minutes, but now we can use this program to quantify plaque from CTA images in five to six seconds."

To create the tool the researchers first trained an algorithm to recognize plaque deposits using a dataset of CTA images from 921 patients. The tool was then validated on a test set of images from several hundred patients, delivering results in close to complete agreement with human expert readers.

Next, the researchers looked at the tool’s capacity to predict future heart attacks. After setting a number of plaque volume thresholds for the tool to work off, the researchers found it could accurately stratify patients into two categories – those at a high-risk and those at a low-risk of experiencing a heart attack within five years of CTA imaging.

“…our study represents the first validation of a deep learning approach for atherosclerotic quantification from CCTA using invasive reference standards, and is the first demonstration of the predictive value of deep learning-based plaque measurements for risk of cardiac events,” the researchers write in the new study.

It is certainly early days for the technology, so don’t expect to find AI doctors giving you heart health advice any time soon. Larger studies will be required to better train the algorithms on diverse patient populations. And even when the technology is optimized there will still be significant hurdles in patient access as CTA imaging is not a cheap or easily accessible diagnostic method.

Nevertheless, the new study is an exciting demonstration of the possible future of medicine. A future where AI tools can swiftly analyze diagnostic imaging to deliver immediate risk reports to patients. Dey is optimistic these kinds of AI tools could be implemented into current clinical workflows to help inform treatment decisions made by doctors and patients.

"More studies are needed, but it's possible we may be able to predict if and how soon a person is likely to have a heart attack based on the amount and composition of the plaque imaged with this standard test," said Dey.

The new study was published in the journal The Lancet Digital Health.

Source: Cedars Sinai