Can we just get our AI act together, please?

4 minute read

The evidence is mounting about the benefits to patients, clinicians, and the health economy, but Australia continues to clutch its pearls.

There are days when the sight of yet another story about artificial intelligence use in healthcare is enough to drive me to the edge of madness.

There is an awful lot of hand-wringing and pearl-clutching being done on this subject.

There have been two government consultations on the responsible use of AI in as many years and so far we seem to be no closer to having a sensible framework for its use in healthcare.

The AMA has released a position statement on the subject, filled with caveats and requests.

“There’s no doubt we are on the cusp of big changes AI can bring to the sector and this will require robust governance and regulation which is appropriate to the healthcare setting and engenders trust in the system,” it says.

“We’d like to see a national governance structure established to advise on policy development around AI in healthcare.

“Such a structure must include all health-sector stakeholders like medical practitioners, patients, AI developers, health informaticians, healthcare administrators and medical defence organisations.

“This will underpin how we carefully introduce AI technology into healthcare. AI tools used in healthcare must be co-designed, developed and tested with patients and medical practitioners and this should be embedded as a standard approach to AI in healthcare.”

Lots of coulda, woulda, shoulda in that lot, which prompted Dr Bertalan Mesko, a Hungarian “medical futurist” of repute to issue a bit of a rebuke.

“I always smile out of disappointment when a medical association ‘calls for stricter regulations on healthcare AI’, just like the Australian Medical Association did,” Dr Mesko wrote.

“It is your job to provide regulations. This is why some other medical associations [Canada and the US] have been working with professional futurist researchers like me to make it happen.” 


The fact is Australia is a laggard on this front, as it is on most matters of digital health reform and – brace yourselves – interoperability. There’s a word begging to be tattooed across the forehead of every health minister for the next 20 years until they get it right.

Professor Enrico Coeira from Macquarie University and colleagues wrote in the MJA recently that dealing with AI’s implications was a “national imperative”.

“With AI’s many opportunities and risks, one would think the national gaze would be firmly fixed on it. [Yet] there is currently no national framework for an AI-ready workforce, overall regulation of safety, industry development, or targeted research investment.

“The policy space is embryonic, with focus mostly on limited safety regulation of AI embedded in clinical devices and avoidance of general-purpose technologies such as ChatGPT.”

And of course the research on AI’s wondrous benefits for the medical profession keeps pouring in.

The latest is a study out of the University of Chicago, published in the Journal of Medical Imaging, in which researchers developed a deep learning-based model that can predict if a patient will need intensive care by analysing their chest X-ray images.

The researchers fine-tuned a large model, pretrained on ImageNet with 1.2 million natural images, using chest X-ray images from a National Institutes of Health dataset to detect 14 different diseases. They then refined this model using a dataset from the Radiological Society of North America to detect pneumonia. Lastly, they fine-tuned it using an in-house dataset, which contained 6685 X-ray images from 3998 patients with covid.

The finished AI model predicted whether a patient with covid would need intensive care within 24, 48, 72, and 96 hours of the chest x-ray with a good degree of accuracy. Patients it identified as being high risk were almost five times as likely to require intensive care, said the authors.

This performance was comparable to similar existing models, even though it relied only on images instead of a combination of images and clinical data.

What does it all mean? According to the authors: “The proposed model … could play an essential role in supporting clinical decision-making and resource management, which in turn would improve the quality of care received by patients.”

Happy days.

If we as a country ever get our act together it could come in handy.

Send story tips in ones and zeroes to

End of content

No more pages to load

Log In Register ×