Bridging Healthcare Gaps with AI
AI tools are increasingly shaping healthcare systems worldwide, augmenting care, streamlining administration, and supporting clinicians with daily workflows. At Adelaide University, these developments are driving research aimed at improving healthcare access, equity, and quality.
Among the University’s initiatives are two world-first projects: Project CANAIRI (Collaboration for Translational Artificial Intelligence Trials) and IBD-Perfect. The projects, both from the Australian Institute for Machine Learning (AIML), are tackling major challenges in AI implementation and patient care.
Project CANAIRI spans 24 academic, industry, and regulatory partners across a plethora of countries and is focused on establishing international guidelines for health systems wanting to evaluate AI tools before implementation. Specifically, the project aims to establish best-practice standards for a form of evaluation known as a ‘translational trial’.
Before introducing an AI tool into a clinical setting, local testing is essential. Without it, the model can produce unsafe, biased, or unreliable outputs that can mislead clinicians and risk patient harm. However, AI tools often move directly from in silico testing – performed entirely within computational environments – to implementation with real patients and data.
Project CANAIRI co-founder and AIML Deputy Director, Associate Professor Melissa McCradden, says a common failing of premature implementation is that the tool simply doesn’t work in the intended setting, wasting time and resources.
“It’s not enough to just build a good algorithm,” she says.
“The closest that many people get to a kind of real-world prospect of performance is that they've trained on a bunch of data.
“And that's not really equivalent to running this tool live, in parallel to clinical care, to assess its clinical performance and thus its value for a health institution.”
A translational trial is a crucial step between in silico development and real-world implementation. Commonly known as a ‘silent’ trial, it evaluates an AI tool within its local setting, but in the background. Outputs are recorded for analysis but are not visible to the clinicians caring for patients and have no direct impact on a patient’s care.
In this sense, the trial functions as a ‘canary in the coal mine’, allowing healthcare systems to assess whether a tool is likely to provide value locally while also identifying and mitigating risks prior to implementation.
Despite growing recognition of translational trials as critical to safe and effective AI implementation, no standardised international guidelines exist for how the trials should be conducted – or whether they should be conducted at all.
Project CANAIRI exists to address this gap.
One of the project’s first priorities was redefining the scope of a translational trial. This necessitated moving beyond the traditional concept of a ‘silent trial’ because of its secretive connotations and its narrow focus on technical evaluation. To be relevant to all healthcare settings, the project team argues, the trials need to encompass a broader set of factors including environmental impacts, cybersecurity risks, human factors, and legal and regulatory requirements.
Professor Carolyn Semmler, lead of Adelaide University’s Applied Cognition and Experimental Psychology Research Group, AIML researcher and committee member of Project CANAIRI, believes these broader factors are fundamental to not only successful, but also responsible implementation.
“One of the unique aspects of Project CANAIRI is to look at the broad, wraparound factors of the sociotechnical system, and the modifications that need to occur within that system, to allow a model to be implemented in a way that’s safe, ethically acceptable, and cost efficient,” she says.
“Those sorts of concerns are what preserve the social contract between people who pay for the healthcare system, but also people who use the healthcare system, and people who work in the healthcare system.”
With their expanded scope in place, the team moved on to an initial round of international surveys and interviews with stakeholders, exploring current practices and key areas of concern. Next, the team will conduct more targeted surveys, interviews, and workshops before potentially bringing key stakeholders from around the world to Adelaide in early 2027 to begin the drafting process.
The finalised guidelines are expected in mid-2027, followed by a period of testing and refinement to ensure their relevance across diverse healthcare contexts, from large academic medical centres to smaller community hospitals.
In the meantime, Associate Professor McCradden plans to release the project’s findings incrementally, including draft guidelines, so that practitioners and researchers have something to work with while they wait.
“I think the thing that has stood out the most to me is that everybody knows this is important,” she says.
“Everybody is struggling with how to do it right, so if we put all these things together for people, you've got a great manual for how to actually do this.”
Alongside these efforts in implementation, Adelaide University is also contributing to the technical development of AI-driven healthcare tools. Since 2024, AIML has collaborated with Crohn’s Colitis Cure (CCCure), an Australian charity dedicated to curing inflammatory bowel disease (IBD), to improve patient care.
IBD encompasses a group of chronic conditions characterised by inflammation of the gastrointestinal tract and debilitating symptoms such as abdominal pain, fatigue, and impaired digestion.
Prevalence of IBD is increasing globally, especially in Western societies. In 2025, almost 180,000 Australians were living with IBD, a figure expected to exceed 200,000 by 2035.
To improve treatment, Professor Lyle Palmer, a geneticist and AIML researcher, has worked with CCCure to develop IBD-PERFECT, the world’s first IBD-specific electronic medical record and patient management platform.
Released at the end of 2025, the platform uses machine learning to generate data-driven insights and analytics from thousands of complex medical records. Specialist clinicians across Australia and New Zealand can access this information to spot trends, identify gaps in care, and plan treatment.
“The platform provides real-time data on clinical quality and performance indicators, as well as informing clinicians about the characteristics of their current patient populations,” Professor Palmer says.
“By connecting patients, clinicians, and researchers, we’re creating a powerful, real-world evidence-base that is already transforming care and accelerating innovation.”
Recently, CCCure joined GLIDE – the Global IBD Registry – an international consortium aiming to establish a global research platform for IBD.
Professor Palmer represents Australia and AIML as co-chair of GLIDE’s Data and Intelligence Committee, which oversees the consortium’s technology strategy. He is currently working with international partners to develop methods for integrating data for advanced machine learning analyses across GLIDE’s international member organisations.
As AI continues to reshape healthcare and society more broadly, Adelaide University will continue to collaborate with global partners to ensure the technology is implemented ethically and equitably, and that research delivers practical solutions to real-world problems.
Adelaide University is Australia’s university for the future ranked in the global top 100.* With more than 180 years of collective experience and achievement, Adelaide University is leveraging the capabilities and resources of its foundation institutions to launch bright futures and breakthroughs.
*Ranked 79, 2027 QS World University Rankings
This content was provided by Adelaide University. The editorial staff of The Chronicle had no role in its preparation. Find out more about paid content.
Built with Shorthand