Modelling has revealed integrating an artificial intelligence (AI) powered scan for diabetic retinopathy into routine healthcare can increase detection of the disease, while also reducing costs across the healthcare system.
The research, led by Dr Wenyi Hu and Associate Professor Lisa Zhuoting Zhu at the Centre for Eye Research Australia (CERA), shows that an accurate AI-powered camera could prevent up to 40,000 cases of blindness from diabetic retinopathy, while also reducing costs to the healthcare system by as much as AU$615 million, over a 40-year period.
By being cheaper to perform than a traditional test and not needing specialised training, an AI scan could be incorporated into many more care settings.
“If by making this test cheaper and able to be introduced into routine care for everyone, we can find and save the sight of people who might not yet know they have diabetes,” Dr Hu said.
“A universal screening program, which is accessible to as many people as possible, could mean more people are able to start treatment before they lose their vision.”
An accurate AI-powered check that is cheaper and easier to perform, while still accurate, could make it possible for the test to be incorporated into many more primary care settings, like general practitioner offices, as a part of routine treatment for everyone.
As well as protecting the sight of people with diabetes, the AI camera could also identify people who have diabetes but have not yet been diagnosed.
“Our modelling has shown that, if we’re able to screen over 80% of the diabetic population to find the undetected cases of diabetic retinopathy in the community and protect their vision, we would not only save their sight but save the healthcare system many thousands of dollars in support services,” Dr Hu said.
“There would be some increase in the cost of treating the disease, but since more people get treated earlier, less people will progress to blindness – which is our ultimate goal.”
The team performed this modelling by comparing an AI-powered system against conventional manual screening at estimated rates of detected and undetected diabetic retinopathy in the community.
The technology requires minimal training, and the portable camera could make healthcare more accessible to regional Australia and remote Aboriginal and Torres Strait Islander communities that might not have regular access to eyecare.
AI scans
AI-powered scans work by taking a picture of the back of the eye where the damage caused by diabetic retinopathy occurs.
The AI system is trained on thousands of images of eyes both healthy and with diabetic retinopathy, and uses this information to look for signs of the disease in patients.
Previous research by the team has found the technology can accurately identify the disease, and research participants have been very positive about their experiences with the AI screening system.
“This research shows that AI screening for diabetic retinopathy can play an important role alongside clinicians in helping people in the community protect their vision,” Dr Hu said.
“AI scans like these have a great potential to increase the number of people we screen for the disease – and working alongside clinicians and policymakers we could one day incorporate this into routine clinical care.”
More reading
AI shown to match or outperform ophthalmologists
AI models predict long-term visual acuity for high myopia
AI in eyecare: myths and attitudes
Reference
Hu, W., Sanil, J., Rui, L., et al. Population impact and cost-effectiveness of artificial intelligence-based diabetic retinopathy screening in people living with diabetes in Australia: a cost effectiveness analysis. eClinicalMedicine (2024). https://doi.org/10.1016/j.eclinm.2023.102387
