A targeted RNA test scans patient samples and spots antibiotic-resistant bacteria in just a few hours — much more quickly than existing clinical tests.
At least 700,000 people a year die from drug-resistant infectious diseases, according to the World Health Organization. Conventional tests for antibiotic resistance involve growing bacteria for a day or more. Newer genetic assays are quicker, but they only detect genes that are already known to make bacteria resistant to drugs.
Deborah Hung at the Broad Institute of MIT and Harvard in Cambridge, Massachusetts, and her colleagues focused on the RNA molecules that bacteria produce when they are exposed to antibiotics. The researchers used machine learning to identify RNA molecules that distinguish drug-resistant bacteria from sensitive strains.
In one test that took under 4 hours, the team’s method successfully determined whether 71 out of 72 samples of Klebsiella pneumoniae and Escherichia coli bacteria were resistant to various antibiotics.