A company that provides hospitals with a treasure trove of medical data to train AI

Artificial intelligence models are ever-hungry black boxes, requiring vast amounts of bits and bytes from wide-ranging real-world data streams to generate insights about patients and their care. To meet this need, a number of companies have emerged that buy patient data from hospitals and sell it to companies interested in AI training and research.

Earlier this week, health data company Truveta, which typically handles data such as patient immunizations, social health decisions, laboratory tests, and pharmacy and insurance claims, launched its new Truveta Genome Project to create a large-scale database of genetic information. announced that it would start. It will target 10 million patients over the next five years and combine their health record data. Companies such as Avandra, Gradient Health, Segmed, and Protege provide anonymized patient images to businesses and researchers.

Although this is legal, Marielle Gross, a bioethicist at Johns Hopkins University, said that the process of capturing people’s data for AI training and scientific discovery is not a good idea because popular processed foods are made from the human body. He likened it to the 1973 Charlton Heston film “Soylent Green,” in which it was revealed that “We’re taking parts of people, whether they want to interpret them digitally or physically, and we’re creating products and selling them to them,” she said.