NIH Will Use AI, X-Rays to Spur Personalized Covid Treatments

Scientists at the NIH are pioneering better Covid-19 diagnostic methods using artificial intelligence and stacks of digital lung and heart images through a new project the agency announced Wednesday.

The National Institutes of Health created a new center that will collect medical images of infected patients and other data in an effort to help doctors tailor Covid-19 treatments to specific people.

With images and data all in one place, specialists will be able to find patterns in how the virus affects the body and create better procedures to treat it. Doctors eventually hope to translate that information into algorithms to determine what type of treatment would best help a patient.

Providers are still waiting for a drug to treat the virus that wins approval from the Food and Drug Administration. Until that happens, providers have been largely on their own in deciding how to treat patients. A diagnostic algorithm for Covid-19 like the one the NIH hopes to create could help bring order to hospitals and peace of mind to providers who aren’t sure how to handle their case loads.

A centralized system would also help confirm various hypotheses about how Covid-19 manifests and how other health issues affect the severity of virus symptoms, said Krishna Kandarpa, director of research sciences and strategic directions within the NIH’s Biomedical Imaging and Bioengineering office, which is leading the project.

“We’re hoping within 3 to 6 months we’d add more ‘oomph’ to what’s out there,” he said in an interview.

Figuring out how far Covid-19 has progressed in a patient is difficult because doctors have to sort through a barrage of different test formats and symptoms, the agency said. The program will create a system to clearly evaluate whether a patient has Covid-19 or something else.

Down the line, scientists hope this sort of project will help providers treat other types of conditions too.

“We’re applying this to Covid-19 because it’s the crisis of the day, but the software technologies and systems behind this will be very relevant later to other respiratory infectious and other organs and diseases,” Kandarpa said.