Sepsis, one of medicine's most time-sensitive killers, has long outpaced the systems designed to catch it — but a team of Australian clinicians may have found a way to close that gap. NSW Health's SAFE-WAIT model uses artificial intelligence to identify sepsis risk in emergency waiting rooms before traditional tests can, flagging vulnerable patients through a simple traffic-light system that guides nurses toward faster intervention. Piloted at Westmead Hospital in 2024, the tool delivered antibiotics to at-risk patients an average of 83 minutes sooner than standard protocols, a margin that in
NSW AI tool cuts sepsis detection time by 83 minutes, expanding to five hospitals
Sepsis is associated with one in five deaths worldwide; early detection via this tool aims to reduce mortality through faster treatment initiation.
Eighty-three minutes earlier than standard care
Mark
So the AI system flags patients at risk of sepsis before traditional tests would catch them. How does it actually know? What's it looking at?
Mimi
It uses the initial assessment data that a triage nurse enters—vital signs, symptoms, patient history. The model has learned patterns from the pilot data that correlate with sepsis risk. It's not diagnosing; it's flagging probability.
Luke
Right, but we should be clear: the source doesn't detail what specific data points the model weighs or how it was trained. We know it works in the pilot, but the mechanism is a bit of a black box from what's reported here.
Mark
Fair. So the 83 minutes—that's the average time saved between when SAFE-WAIT flags someone and when they'd normally get antibiotics?
Mimi
Yes. During the Westmead pilot, patients the model identified received antibiotics 83 minutes earlier than the standard pathway would have given them.
Luke
That's a significant finding, but it's from one hospital over six months. The expansion will test whether that holds across five different settings. That's the real question.
Mark
Why does it matter that it's expanding to five hospitals instead of just rolling out everywhere?
Mimi
Because sepsis detection and patient flow are different at every hospital. Different staffing, different populations, different workflows. If the model only works at Westmead, it's not a solution—it's a curiosity.
Luke
And we don't yet know if it will work at those other sites. The expansion is designed to answer that question. The funding covers 2.5 years of evaluation.
Mark
What's the actual human stakes here? Why should someone care about this beyond the hospital system?
Mimi
Sepsis kills one in five people globally. It's fast and it's brutal. If this tool can reliably catch it earlier, it could prevent deaths. That's not abstract.
Luke
True, but we should note: the source doesn't give us mortality data from the Westmead pilot. We know patients got antibiotics faster, which is good, but we don't have hard numbers on whether that translated to fewer deaths or better outcomes.
Mark
So faster treatment is the theory, but the proof is still being gathered.
Mimi
Exactly. The pilot proved the model could identify risk earlier. The expansion will test whether that early identification actually improves patient outcomes across different hospitals.
El Pulso
- Sepsis kills one in five people globally, and its lethality is inseparable from speed — every minute without treatment narrows the window for survival.
- Emergency waiting rooms are the blind spot: patients whose conditions are deteriorating sit unseen, and conventional diagnostics are too slow to catch sepsis before it escalates.
- SAFE-WAIT's traffic-light AI flagged high-risk patients at Westmead Hospital 83 minutes earlier than standard methods, a result striking enough to secure $500,000 in government grant funding.
- The system is designed not to replace clinical judgment but to augment it — giving nurses a continuously vigilant second set of eyes in the controlled chaos of a busy ED.
- Beginning in early 2027, the model rolls out to four additional Sydney hospitals over two and a half years, with its true test being whether the 83-minute advantage holds across varied settings, staffing, and patient populations.
Sepsis, one of medicine's most time-sensitive killers, has long outpaced the systems designed to catch it — but a team of Australian clinicians may have found a way to close that gap. NSW Health's SAFE-WAIT model uses artificial intelligence to identify sepsis risk in emergency waiting rooms before traditional tests can, flagging vulnerable patients through a simple traffic-light system that guides nurses toward faster intervention. Piloted at Westmead Hospital in 2024, the tool delivered antibiotics to at-risk patients an average of 83 minutes sooner than standard protocols, a margin that in sepsis can determine whether a person walks out of hospital or does not. Backed by nearly $500,000 in translational research funding, the model will now expand across five hospitals — a test of whether a promising idea can become a durable, life-saving system.
Sepsis is a condition defined by time. The body's immune response turns catastrophic, organs begin to fail, and the hours lost to delayed diagnosis can be irreversible. It is responsible for one in five deaths worldwide — and yet it frequently goes undetected in the very place patients first arrive for help: the emergency department waiting room.
NSW Health's SAFE-WAIT project was built to address exactly that gap. Developed by a team of clinicians led by Associate Professor Dr Amith Shetty, an emergency physician at Westmead Hospital, the AI model integrates with triage systems to analyze initial patient data and assign a risk level — green, amber, or red — before conventional diagnostic tests would raise any alarm. The design is deliberately simple, suited to the noise and urgency of a working emergency department.
A six-month pilot at Westmead in 2024 produced a striking result: patients identified by SAFE-WAIT received antibiotics an average of 83 minutes earlier than those assessed through standard protocols. That figure was compelling enough to attract nearly $500,000 through the Translational Research Grants Scheme, funding an expansion to Auburn, Blacktown, Mount Druitt, and Nepean Hospitals alongside Westmead, beginning in early 2027.
Shetty frames the tool not as a replacement for clinical expertise but as a way to sharpen it — helping clinicians see what might otherwise be missed in a crowded waiting room. NSW Health Ministers Ryan Park and David Harris have both pointed to the expansion as evidence of the state's commitment to moving research out of pilots and into the hospitals where it can change outcomes.
The pilot was proof of concept. The expansion across five hospitals, with different patient populations and institutional cultures, is the real test — of whether 83 minutes can become not a single-site achievement, but a repeatable, scalable margin between life and death.
Sepsis kills one in five people worldwide. It arrives without warning—the body's immune system spiraling into catastrophic overdrive in response to an infection, shutting down organs, stealing hours that might have meant survival. The difference between life and death often comes down to minutes. A team of clinicians at NSW Health has built an artificial intelligence system designed to steal those minutes back.
The SAFE-WAIT project is an AI prediction model that sits inside emergency department waiting rooms, watching. When a triage nurse enters initial assessment data into the system, the model analyzes it and flags patients at risk of developing sepsis before traditional diagnostic tests would catch it. The system works like a traffic light: green for low risk, amber for moderate, red for high. That color tells staff whether to escalate care immediately or monitor more closely. It is simple, visual, and built for the chaos of an emergency department where dozens of patients are waiting and time is the one resource that cannot be recovered.
The proof came first at Westmead Hospital. During a six-month pilot in 2024, the model showed what early detection could do. Patients flagged by SAFE-WAIT received antibiotics an average of 83 minutes earlier than they would have under standard protocols. Eighty-three minutes. In sepsis, that gap between waiting and treatment can be the difference between discharge and intensive care, between survival and the alternative. The pilot worked. The results were clear enough that the project earned nearly $500,000 in grant funding through the Translational Research Grants Scheme, money that will support expansion across five hospitals over the next two and a half years.
The team behind SAFE-WAIT is led by Associate Professor Dr Amith Shetty, an emergency physician who understands the particular vulnerability of patients sitting in waiting rooms—people who have not yet been seen by a doctor, whose condition may be deteriorating in plain sight. The tool is designed to catch them before they crash. Shetty describes it as helping clinicians "see the full picture," identifying sepsis much earlier than conventional methods allow. It is not meant to replace clinical judgment. It is meant to sharpen it, to give nurses and doctors a second set of eyes that never blinks and never forgets that sepsis waits for no one.
Starting in early 2027, the model will expand to Auburn, Blacktown, Mount Druitt, and Nepean Hospitals alongside Westmead. The rollout will take two and a half years, and the evaluation will test whether the system can work reliably across different hospital settings, different patient populations, different workflows. That is the real test—whether something that worked at one hospital can work at five. Whether the promise of 83 minutes can hold up when the variables multiply.
NSW Health has positioned this expansion as part of a broader commitment to artificial intelligence in healthcare. The state has been a national leader in sepsis recognition and management since 2011, and this tool represents the next evolution: not just teaching clinicians to spot the disease faster, but building systems that do some of that spotting automatically. Minister for Health Ryan Park called it "cutting-edge technology," and Minister for Medical Research David Harris framed it as part of a larger mission to translate research into real-world care—to take something that works in a pilot and make it work in the hospitals where people actually get sick.
What happens next depends on whether the model can survive contact with reality at scale. The pilot was proof of concept. The expansion is the test. If SAFE-WAIT can maintain its performance across five different emergency departments, with different staffing levels and patient mixes and institutional cultures, then the 83-minute advantage becomes not a one-hospital story but a template. And in a condition where every minute counts, a template that saves time could save lives.
Citas Notables
Within the pilot we undertook we tracked that on average patients were administered antibiotics 83 minutes earlier than without the model, showing how proactive it is.— Associate Professor Dr Amith Shetty, emergency physician and SAFE-WAIT project lead
With sepsis, every minute counts. This investment in AI technology will help clinicians identify it early, improving the safety of patients in ED waiting rooms.— Minister for Health Ryan Park
Quieres la nota completa?
Lee el original en The National Tribune ↗
Technology