New South Wales will spread an AI sepsis detection model across five hospital emergency departments after a pilot at Westmead Hospital showed it could flag high-risk patients faster than standard testing.
The SAFE-WAIT system reads triage data entered by a nurse and sorts patients into low, moderate or high risk of sepsis, working much like a traffic light. Clinicians use the signal to escalate care sooner, while blood cultures and other confirmatory tests are still pending.
The project received almost A$500K through the state’s Translational Research Grants Scheme, part of a A$4.9M package for high-impact research announced on World Sepsis Day. The rollout covers Westmead, Auburn, Blacktown, Mount Druitt and Nepean hospitals over two and a half years, with work resuming in early 2027.
Emergency physician Associate Professor Amith Shetty leads the clinical team behind the model. The next phase will test whether the algorithm holds its accuracy when it moves between sites with different case mixes and staffing.
Sepsis outcomes depend heavily on time to treatment, which makes the waiting room an attractive target for prediction tools. Several US health systems have deployed similar sepsis algorithms, and researchers have warned that some models can be skewed by how their training data records the moment of onset.
NSW Health also convened an AI forum this week to examine how such tools fit into clinical workflows and governance. No decision has been announced on statewide adoption beyond the five-site evaluation.
