Companies pay hospitals for health data sets to train artificial intelligence

AI models are ever-hungry black boxes that need lots of bits and bytes from a wide stream of real-world data to gain insight into patients and their care. To meet this need, a slew of companies have sprung up to buy patient data from hospitals and sell it to people looking to train artificial intelligence or conduct research.

Earlier this week, health data company Truveta, which typically trades in patient vaccination data, social health assessments, laboratory tests, and pharmaceutical and insurance claims, announced that it would begin new Truveta genome project creating a massive database of genetic information from 10 million patients over the next five years to link it to their health data. Many companies, including Avandra, Gradient Health, Segmed and Protege, offer de-identified patient images to companies and researchers.

Although it is legal, Johns Hopkins bioethicist Marielle Gross likened the process of obtaining data from humans to train artificial intelligence or make scientific discoveries Green solventCharleton Heston’s 1973 film in which a popular processed food product is revealed to be composed of human bodies. “We take pieces of people — whether you want to interpret it digitally or physically — and cut them up, create products, and then sell them to them,” she said.

STAT+ exclusive story

STAT+

This article is intended for STAT+ subscribers only

Unlock this article—and get additional analysis on the technologies disrupting health care—by subscribing to STAT+.

Already have an account? Log in

View all plans

To read the rest of this story, subscribe to STAT+.