8 July 2019

The deadly perils of not investing in medical AI

KnowCents Technology

“We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.”

– Amara’s Law

Artificial Intelligence, or AI, has a public relations problem in Australia. It’s been coming so long, with such hype and promise, and failed to deliver so spectacularly – IBM Watson and cancer diagnosis, as the most recent example – many in the sector aren’t really sure what is going on.

Last month, one of our most recognised experts in the field of medical AI, Professor Enrico Coiera, The Director of the Australian Institute of Health Informatics, made the situation clear. Medical AI is here and gathering momentum, and unless we organise ourselves faster for this change we could miss out on significant opportunities locally, or worse, end up with AI medical failures through poor governance and education.

Professor Coiera said Australia would never likely be a hub for serious medical AI development, given the march of the major global digital distribution platforms such as Google, Apple and Amazon, but that the Australian healthcare ecosystem was an ideal testing ground for major overseas investors and this was a way in for this nation to participate meaningfully in this revolution.

He pointed to the benefits Australian patients and researcher gain from our significant involvement in the global clinical drug trials market.

“It’s [medical AI] real. We are in a world that has already changed,” Coiera told a Medical Software Industry Association gathering in Sydney last month, citing the massive investments now being made in medical AI in the UK and the US, both privately and by government.

Britain’s National Health Service had allocated £1 billion to medical AI research and implementation and set itself the goal of cutting cancer deaths by 10% (22,000 lives) per year by 2035. And that the same potential was now there for Australia to improve health and save lives, but so far there had been very little traction to fund or organise medical AI, Professor Coiera said.

In the past 18 months, Professor Coiera has been working hard at establishing an Australian body to oversee governance and R&D for our local AI community, called The Australian Alliance for Artificial Intelligence in Healthcare (AAAIH).

The key goals of the organisation include supporting the acceleration and adoption of existing AI-enabled health services (of which there are quite a lot now), building a community of shared interest and pooled resource, undertaking targeted local research, building a national workforce and possibly, most importantly, ensuring that medical AI here is safe and ethical.

Professor Coiera said we were going to need to formulate a national approach to AI in medicine if we wanted our local sector to be safe and have ethical governance.

“In Australia today, there are few standards or governance to oversee safe and effective use of digital health and there are none specifically dealing with AI”, he told the MSIA members.

Another key to the new organisation is making sure Australia has close ties with appropriate international groups.

In its first 18 months, the AAAIH has signed up 52 formal partners, including five state health departments, key AI industry players, such as Google Health, Accenture, Amazon, IBM and Oracle, many of our major universities, our key related digital health associations, such as the RACGP, and frmed partnerships with the universities of Stanford, Oxford, Toronto, Michigan and Liverpool.

Professor Coiera said government was starting to take notice, particularly at the state level, but that given the pace of change in the US and the UK, the sector needed more attention, and a lot more funding. And soon.