Skip to Main Content

Medical datasets marketed by data aggregators contain detailed information about the health conditions of millions of Americans. With the help of artificial intelligence, they could also be a boon in efforts by both providers and drug makers to identify people who don’t yet realize they have a serious disease.

In a study published Monday, a machine learning model scoured aggregated medical data on 170 million Americans and flagged 1.3 million of them as likely to have an inherited condition known as familial hypercholesterolemia, which causes high cholesterol and elevates risk for heart attacks and strokes. It’s not clear how many of the people who were identified actually have FH. But the project has spurred a first-of-its-kind effort to partner with health systems to notify patients identified by the AI system, and, when appropriate, get them diagnosed and treated.


The approach, led by the nonprofit FH Foundation, could also provide a template for drug makers to identify new consumers for treatments. Drug companies have long funded screening programs, like sponsorships at community health fair booths or genetic testing for people suspected of having an inherited disease. But by mining medical data with AI, they could conceivably turn up far more patients. 

Unlock this article by subscribing to STAT+ and enjoy your first 30 days free!