Young children in hospitals in Western Australia are set to benefit from innovations in artificial intelligence thanks to collaboration between the Pawsey Supercomputing Research Centre and Child and Adolescent Health Service. Healthcare practitioners face numerous pressing challenges, including workforce shortages and overwhelming volume of medical information.
Generative AI offers potential solutions to these issues by streamlining administrative tasks, improving access to information, and ultimately enhancing overall patient care.
Prompt replies to patient enquiries
In 2022 researchers at The University of Western Australia and the Perth Children's Hospital, Dr Thomas Drake-Brockman and Dr Harry Smallbone, started by testing a program in which parents received conversational SMS messages to ask about their child’s post-operative recovery. Both Dr Drake-Brockman and Dr Smallbone are dual-trained medical doctors and computer scientists, who used this intersection of skills to tackle this novel work. Although the trial was focused on data collection methods, as a positive side effect parents expressed a greater sense of connectedness with the hospital. The pair believes that further advances in generative AI are highly applicable to medicine and could help improve medical care.
Dr Smallbone is working towards a PhD, investigating how AI can improve the perioperative anaesthetic journey for patients. Taking the previous SMS trial to a new level, technology could create responses or explanations for parents when complications happen, which would then be checked by a human with access to all the patient’s notes. While the chatbot would never interact directly with a patient or their parents, it could greatly speed up the process of getting the right information from a doctor.
For example, if a parent was in touch with the hospital about their child, who has been home for two days but now has a fever and would like to know what to do, rather than a doctor reading through all of the notes, guidelines and recommendations, an AI program could summarise the notes, extract relevant information, and suggest a way forward.
Dr Drake-Brockman and Dr Smallbone emphasise that it is vital to use clinician-in-the-loop methods – where the human expert has the final say. With an AI making suggestions, it is crucial that there is no incentive toward the advice given: “We have to make sure that there is no more work associated with rejecting the advice than accepting it and passing it on to the patient. You don’t want to incentivise clinicians to just take the advice and move on to the next bit of work.”
Dr. Thomas Drake-Brockman (left) and Dr. Harry Smallbone (right) are leveraging the power of Setonix to improve pediatric care.
Enhancing wellbeing and recovery
Their team is led by the newly announced Life Scientist of the Year Professor Britta Regli-von Ungern-Sternberg, Chair of Paediatric Anaesthesia at The University of Western Australia, Director of the Institute for Paediatric Perioperative Excellence, and a specialist paediatric anaesthetist at Perth Children’s Hospital. She took the 2024 Prime Minister’s Prize for Science award for work that has significantly reduced risks of anaesthesia in young children and transformed clinical guidelines and changed medical practice around the world.
“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. Well-trained, human experts verified AI has great potential to support clinicians in the future to give the best care to every patient,” said Prof Regli-von Ungern-Sternberg.
The team upholds high standards in ethical management of data, and as such are focused on how to bring the processing power of Pawsey’s Setonix, the greenest and most powerful supercomputing in the southern hemisphere located in Perth, to help address challenges in the field of child and adolescent medicine. Setonix can be used to train algorithms using open access and synthetic data while actual patient data remains securely in the hospital.
Dr Drake-Brockman says: “There is a huge wave of administrative work in medicine, and at the same time, not enough doctors to respond to patients’ queries. Large language models offer a lot of promise to tackle some of these challenges, but the data analysis techniques are still experimental. With improved data flow and management, and human oversight, we hope we will soon have tools to support medics and in turn, improve patient safety.”
The team are also bringing processing power closer to the bedside. A new system dubbed MERLIN has recently been installed by the team within Perth Children's Hospital, in partnership with The University of Western Australia, to support ongoing work in AI and data analysis. Generously funded by the Stan Perron Charitable Foundation, MERLIN will ensure patient data remains within a secured health IT environment when advances in AI are used to enhance care, and will allow work done on Pawsey's Setonix to be translated into practice at the clinical coalface.
The synergy between technology and medicine
Looking to the future, they are considering how robots could enhance pre-operative assessments for children. Dr Smallbone explains: “We noticed in other studies that some kids can find it easier to interact with a friendly robot than with a human, particularly when they are nervous or anxious. So we are exploring ways to use AI models that are capable of language generation that could understand what a child or their parents are saying and then produce a text or even a spoken response.”
They see a growing need for computer science in health research and are excited to work with others to try out new and experimental data analytics techniques.
