Researchers at Johns Hopkins University deployed a machine learning early warning system called TREWS across five hospitals, watching over 590,000 patients in real time and flagging signs of sepsis before a clinician might otherwise notice them.
When a provider confirmed a TREWS alert within three hours, in-hospital mortality fell by an adjusted 18.7 percent relative to patients whose alert went unconfirmed in that window, and median time to a patient's first antibiotic dose dropped by close to two hours.
The benefit depended entirely on a human acting on the alert quickly; a flag nobody confirmed changed nothing about outcomes. This was an observational comparison, not a randomized trial. Some alerts were false positives that added to clinician workload, and adoption varied by hospital and by provider, a factor the research team studied separately as its own problem. The system's success came from pairing it tightly with how fast staff actually responded, not the algorithm by itself.
Read the original study
Adams et al., Nature Medicine, 2022 · doi.org/10.1038/s41591-022-01894-0
An earlier alarm, and almost a fifth fewer deaths.
Source
- Adams, R., Henry, K. E., Sridharan, A., Soleimani, H., Zhan, A., Rawat, N., Johnson, L., Hager, D. N., Cosgrove, S. E., Markowski, A., Klein, E. Y., Chen, E. S., Saheed, M. O., Henley, M., Miranda, S., Houston, K., Linton, R. C., Ahluwalia, A. R., Wu, A. W., & Saria, S. (2022). Prospective, multi-site study of patient outcomes after implementation of the TREWS machine learning-based early warning system for sepsis. Nature Medicine, 28(7), 1455–1460. https://doi.org/10.1038/s41591-022-01894-0