AI predicts how likely you are to have a heart attack in the next 5 years

A new AI tool will make it easier to predict heart attacks in people. Model described in the Lancet

Digital Health accurately predicts which patients will have a heart attack in five years based on the amount and composition of plaque in their arteries.

Plaque buildup can lead to narrowingarteries, making it difficult for blood to reach the heart and increasing the chance of a heart attack. A medical test called a coronary computed tomography angiography (CTA) provides 3D images of the heart and arteries to see how narrowed they are. But until now, there hasn't been a simple, automated, and quick way to measure the plaques that are found on CTA images.

Therefore, the authors decided to create a working model,which will do it quickly and efficiently. First, they trained an algorithm to recognize plaque deposits using a CT angiography image dataset from 921 patients. The tool was then tested on a test set of images of several hundred patients. AI conclusions almost completely coincided with the opinions of expert doctors.

The expert needs at least 25-30 minutes, but the program does the same in five or six seconds!

Damini Dey, laboratory director at the Biomedical Imaging Institute at Cedars-Sinai

Next, the authors trained the algorithm to predictfuture heart attacks: They identified a number of plaque volume thresholds. As a result, the tool could accurately classify patients into two categories—high-risk and low-risk for heart attack within five years of CT angiography.

Of course, more research is needed. But we may already be able to qualitatively predict whether and how soon a person will have a heart attack. 

Damini Dey, laboratory director at the Biomedical Imaging Institute at Cedars-Sinai

Read more:

"James Webb" took the clearest photo of a star in history

Quantum charging will allow record-breaking fast charging of electric vehicles

NASA photographed the Chinese Zhurong rover from orbit