An AI-enabled healthcare sector is a potentially idyllic place, where healthy habits are supported, early disease is detected, and, ultimately, deaths are prevented.
But while 85% of healthcare leaders have an AI strategy, much of this multibillion-dollar industry remains untapped, with clinical AI virtually non-existent in Australian hospitals.
Of those who are dabbling in other AI uses, penetration is limited, with only half of healthcare providers currently using the technology in other forms.
This contrasts with sectors like financial services, where global AI spend is projected to reach $97 billion by 2027.
So what is impeding the healthcare industry’s uptake of this readily available technology?
Access to data
The quality of AI decision-making from large language models largely rests on the quality of data used to train it — a realisation that has helped earn data kudos as the ‘new oil’.
Analysing data from electronic health records, an award-winning AI tool by Telstra Health, RMIT and the Digital Health CRC, for example, can detect early signs of deterioration in frail aged care residents.
“It monitors structured and free-text EHRs for 36 evidence-based signs of deterioration. In turn, it provides staff with a frailty index for each resident, and alerts them to falls, depression and mortality risk,” said Annette Schmiede, CEO of Digital CRC, the organisation that facilitated the research initiative.
However, as Schmiede points out, accessing healthcare data continues to be challenging, with organisations still facing delays when requesting it.
“We are still experiencing significant delays in access to data, despite it being constantly identified as an important element in driving innovation,” she told Hospital + Healthcare.
“An example of this is a project we recently ran, which can only commence after two years of waiting for the data to be made available by the state authority.
“We have excellent relationships with the data custodians, we’re well respected, but the whole process is still taking too much time — and that needs some urgent attention.”
While Schmiede is encouraged by the government’s work with health information exchanges — in which health data is being uploaded onto common platforms — she says more work is needed to improve access in the interim.
“There’s widespread acknowledgement of the issue, but we’re still not seeing that being translated into faster access.”
Interoperability
The seamless exchange of data between disparate parts of the healthcare system is also important for end-users — and an ongoing source of frustration for those using new technologies to track health metrics.
“It’s great if you have an AI-powered health app that monitors your vitals over time. But if that information is not automatically added to your health records, then you need to repeatedly explain it to your GP, then to a specialist, and to an allied health professional.
“It affects the user experience for staff and patients, and opens the door for errors and omissions, which takes away a lot of the benefit,” Schmiede said.
According to Dr Stephanie Allen from Kearney, part of the interoperability puzzle is ensuring that the data and insights collected by personal health devices or apps, are “clinical grade”, and don’t produce false positives.
“False positives can create unnecessary anxiety for users but also swamp the already stretched primary health system, with the expectation to ‘test’ again before a formal diagnosis is given,” she said.
In addition, data from different apps and devices should be brought together to paint a full picture of a person’s health.
“We know that our health is interconnected with many aspects of our lifestyle. For example, the relationship between mental wellbeing, nutrition and sleep is only beginning to be understood.
“Combining this data to form a more holistic picture of an individual’s health is fundamental to making the right behavioural adaptations, to protect and/or enhance our overall wellbeing.
“If we have variable levels of quality and reliability of data capture this becomes an impossibility,” Allen said.
Upskilling the workforce
Even a well-oiled healthcare system, with seamless connectivity, will not support the use and uptake of AI if end-users cannot easily navigate it.
For this reason, Digital CRC is taking steps to improve digital literacy and upskill the healthcare workforce, through its Education and Capacity building and Emerging Leaders programs.
“These were set up with the aim of producing the next generation of digital health professionals. Through this program, they will learn first-hand how to use digital health and data analytics to improve patient outcomes and clinician experiences,” Schmiede said.
“We are empowering this group to revolutionise health care and become change agents.”
Optimistic outlook
With projects like these, Schmiede is confident that Australia’s AI future is bright and that more people will feel empowered to use the technology.
“By introducing it responsibly, we can gradually build industry confidence. And as AI becomes more widespread, we’ll be forced to address issues like data access and interoperability head-on,” she concluded.
Image credit: iStock.com/ArtemisDiana
