By Liz McGrath
The collaboration between UWA researchers, Perth Children's Hospital and WA's supercomputing facility is revolutionising patient care through ethical AI implementation.
Sometime in the near future, a parent at home with a child recently discharged after surgery from Perth Children's Hospital may receive a personalised message checking on their post-operative recovery.
While drafted by artificial intelligence, the response will have been verified by a clinician – part of a new system being developed through a partnership between UWA, Perth Children's Hospital and the Pawsey Supercomputing Research Centre, all located in Perth.
The collaboration, led by 2024 Prime Minister's Prize for Science winner Professor Britta Regli-von Ungern-Sternberg, alongside Dr Thomas Drake-Brockman and Dr Harry Smallbone, is harnessing WA's extraordinary computing power to address pressing healthcare challenges.
"In the perioperative environment, clinicians often have to consider and interpret a large amount of information, sometimes in acute situations, with very little available time," says Professor Regli-von Ungern-Sternberg, a consultant anaesthetist at PCH, Chair of Paediatric Anaesthesia at UWA, and co-leader of the Perioperative Care Program and Perioperative Medicine Team at The Kids Research Institute Australia.
Image: 2024 Prime Minister's Prize for Science winner Professor Britta Regli-von Ungern-Sternberg, (centre), with Dr Thomas Drake-Brockman (left) and Dr Harry Smallbone (right).
"Well-trained, human expert-verified AI has great potential to support clinicians in giving the best care to every patient, allowing for more detailed and faster responses to individual queries."
At the heart of this technological advancement are two complementary systems. Setonix, Australia's most powerful supercomputer housed at the Pawsey Centre, trains complex AI algorithms using synthetic data.
Meanwhile, MERLIN, a new system generously funded by the Stan Perron Charitable Foundation and installed within PCH's secure environment, refines the models and will allow the innovations to be safely applied to real patient data in the future.
Clinical trials will take place before the new system is rolled out to routine care, always prioritising patient safety and data security, says Professor Regli-von Ungern-Sternberg.
"It's about the synergy between technology and medicine," explains Aditi Subramanya, Manager of Partner Engagement at Pawsey. "The AI isn't making independent decisions about patient care. It's taking on the administrative burden by processing vast amounts of information and making recommendations that clinicians review before communicating with patients."
The practical applications are promising. In the future, when a parent may contact the hospital about their child's post-operative symptoms, the system could instantly analyse patient history, relevant medical guidelines and similar cases to suggest appropriate responses – which will be verified by medical professionals before being shared. This will ensure that parents receive timely responses and allow clinicians to efficiently manage patient care.
"What makes this approach special is that it's clinician-in-the-loop," says Dr Drake-Brockman. "The human expert always has the final say. We've designed the system so there's no incentive to simply accept the AI's recommendation without proper review."
Ms Subramanya says the environmental considerations of running a computing colossus like Setonix haven't been overlooked with groundbreaking geothermal cooling technology employed, using water from an aquifer beneath the Pawsey facility, to manage the intense heat generated by its processing power.
"When we were looking to procure the new system, minimising our environmental impact was very important," says Ms Subramanya, who describes her role at Pawsey as a dream job. "Setonix represents our commitment to both advancing science and protecting our environment."
Another direction for the project includes exploring how robots might assist in pre-operative assessments for children. Dr Smallbone's research suggests some children find it easier to interact with a friendly robot than with humans when feeling nervous or anxious.
Ms Subramanya says by bringing together medical expertise, computer science and ethical frameworks, researchers are creating solutions that can not only ease pressure on an overloaded health system but also improve patient experiences and outcomes.
"It's about enhancing human capabilities, not replacing them," she says. "We're taking the grind out while ensuring patients receive excellent care."
Read the full issue of the Winter 2025 edition of Uniview [Accessible PDF].
