The problem: If someone with chest pain gets a computed tomography (CT) scan that shows no immediate danger, they might not be out of the woods — in the U.K., roughly two-thirds of people who get sent home eventually have a major cardiac event, like a heart attack.
The solution: Researchers developed an AI that can detect risk factors for heart problems that humans miss. CT scans contain a lot of data that the human eye can’t see, like the characteristics of blockages, or if the fat surrounding the arteries has changed, suggesting inflammation. But those are the things the AI can detect after boosting the image quality, especially in the “cardiac shadow,” a part of the scan that looks like a silhouette to humans.
What’s next: The AI is being evaluated for inclusion in the U.K.’s national health system, as well as by the FDA in the U.S. The team behind the project is also exploring whether similar AIs could be used to predict stroke and diabetes.