AI spots potential COVID-19 treatment drug within two days

Jackie Drees -

London-based artificial intelligence startup Benevolent AI used automated language tools to comb through an interconnected database of coronavirus-related biological processes and scientific research to identify a drug that might possibly treat COVID-19, according to The New York Times.

The AI tools can teach themselves to understand written and spoken language by analyzing thousands of books, articles and digital text. Over two days, Dr. Peter Richardson, vice president of pharmacology at BenevolentAI, and his research team used the data findings from the tools to map out connections between particular human genes and biological processes affected by the coronavirus. The team came up with two genes, which Dr. Richardson then created a type of digital flow chart to visualize how existing medications targeted the genes.

Dr. Richardson and his team pinpointed baricitinib, an Eli Lilly drug designed to treat rheumatoid arthritis. BenevolentAI discovered that baricitinib might also prevent the COVID-19 infection itself by blocking the way the virus enters the cells. The AI company told the NYT that it had no expectations for making money from its research and did not have a prior relationship with Eli Lilly.

The Eli Lilly drug has a warning label that it can make infections worse, but it works by suppressing the immune system, which researchers think could suppress a cytokine storm, or an immune system overreaction seen in some severely ill COVID-19 patients.

The drug will be tested in an accelerated clinical trial with the National Institutes of Health, but health officials warn that baricitinib should not be used to treat patients until the results of the trial are available, which are expected in a few months. Even then, it is still not clear how the drug will impact coronavirus patients, said Dr. Dan Skoyronsky, chief scientific officer at Eli Lilly, told the publication.

Copyright © 2024 Becker's Healthcare. All Rights Reserved. Privacy Policy. Cookie Policy. Linking and Reprinting Policy.