AI diabetes screening pilot launches in South Auckland

NEWS - eHealthNews.nz editor Rebecca McBeth 

An artificial intelligence (AI) diabetes retinal screening pilot programme aimed at improving screening rates for Pacific community in South Auckland has been launched by Health Minister Simeon Brown.

The pilot programme involves trained community workers from seven general practices in South Auckland taking retinal images of patients living with diabetes, including those who may be waiting for a hospital appointment.

Images are screened by an AI tool, with a back-up review by an ophthalmologist.

It is already being trialled at Baderdrive Doctors as well as Southpoint Family Doctors and a further five providers will progressively roll-out the pilot.

Across the seven providers involved, each will screen 10 consenting patients a week. Over the three-month pilot period, about 840 patients are expected to be involved.

Harriet Pauga, regional director northern region, Pacific Health, Health New Zealand | Te Whatu Ora says the pilot period will provide Health NZ with the proof of concept to inform the next phase of the AI component, which requires a larger volume of patients (at least 3,000) to validate.

“If the technology is proven through the pilot, it has the potential to be used with a wider scope in primary care to reduce waitlists in secondary services, support consumers to better engage in their care pathways, and inform future national developments of AI in diabetes retinal screening (DRS) she says.

Diabetic retinopathy is a condition that can be caused by diabetes and can lead to impaired or lost vision.
It often has no symptoms in the early stages, which is why early detection is so important, but there are over 26,000 people in South Auckland who have not had their recommended screening in the last two years and Pacific peoples make up 41 per cent of this group.

Pauga says that for Pacific people type-2 diabetes is one of the leading causes of the seven-year life expectancy gap in the Northern Region, and Pacific people have higher prevalence and complications as a result.

Minister Brown says the aim of the pilot is to reduce the risk of people with diabetes developing sight-threatening disease.

“The use of AI in the primary grading of diabetes retinal screening images is expected to increase access to DRS through enabling real-time screening results,” Brown says.

“Real-time screening results then enable the processing of large volumes of data quickly and efficiently.
Brown says the pilot will shorten wait times for patients and increase screening volumes, while freeing up hospital specialists to attend to core clinical practices.

Director of the AI Lab at Health New Zealand | Te Whatu Ora, CK Jin, told the HiNZ AI in Health Workshop in December 2024 that the organisation is testing the use of AI in a semi-autonomous model, where the AI acts as a primary grader.

The focus of the pilot is on AI’s ability to differentiate normal from abnormal cases and mild from severe disease.

“We think that by integrating AI into acting as a primary grader, we are able to keep the current level of sensitivity specificity, whilst reducing quite a large burden of the workload,” he said.

External validation efforts included testing two commercial AI tools on 4,000 historical images to benchmark performance.

The Minister's office said the name of AI tool being used in the pilot cannot be disclosed due to commercial sensitivity.

Note: this image used in this article was made using AI

You’ve read this article for free, but good journalism takes time and resource to produce. Please consider supporting eHealthNews by becoming a member of HiNZ, for just $17 a month

Read more Clinical Software news

Return to eHealthNews.nz home page