Blood clots, leaking blood vessels, and organ damage all are possible outcomes of sepsis, the body’s extreme reaction to an infection. Each year in the United States, 1.7 million U.S. adults develop sepsis, and at least 350,000 either die in the hospital or are sent to hospice because of it.

Sepsis survival depends on rapid detection. That’s problematic because there is no one test for sepsis, and its symptoms mimic those of many other conditions.

Now, an artificial intelligence driven program developed at Johns Hopkins Medicine detects sepsis nearly six hours sooner on average than traditional methods.

“Hundreds of lives likely have been saved by the system,” says Albert Wu, MD, director of the Johns Hopkins Center for Health Services and Outcomes Research.

To create the process, called TREWS — Targeted Real-time Early Warning System — experts fed the AI algorithm thousands of previous patients’ health records so it could recognize signs of sepsis, explains Suchi Saria, PhD, director of the Machine Learning, AI and Healthcare Lab at Johns Hopkins.