SERI researchers develop AI tools for screening kidney disease, predicting age using eye photos
Researchers from the Singapore Eye Research Institute have developed two novel AI-powered tools for screening chronic kidney disease and predicting a person’s biological age from retina photos.
Both tools use AI deep learning algorithms to scan photos of an individual’s retina to assess their health condition.
WHAT THEY DO
The kidney disease screening tool called RetiKid was developed by SERI and the National University of Singapore’s School of Computing in 2019. The tool was trained with over 23,000 retinal images from nearly 12,000 participants in Singapore and China.
In a study, it was shown to have an accuracy of 91% in an internal test and 73% and 83% accuracy in two external test sets. The screening tool has been licensed to health tech startup EyRIS for production and commercialisation.
Both retina and kidneys share a “close biological relationship,” according to Charumathi Sabanayagam, deputy head of SERI’s Ocular Epidemiology Research Group. ”Thus, problems with blood vessels in the retina could provide clues to changes in kidney blood vessels”.
RetiKid may be used as a preliminary screening test for detecting CKD in general populations and high-risk groups, such as those with diabetes. Once done, patients can be counselled to proceed with routine confirmatory blood and urine tests.
The other AI tool, RetiAge, was later developed in 2021 by SERI and Medi Whale, a healthcare startup in South Korea. It was tested in more than 129,000 retinal images from over 40,000 South Korean individuals. The tool was also evaluated for its ability to predict the 10-year risk of disease and death among 56,000 persons in the UK Biobank database.
The blood vessels in a person’s retina too can show their ageing process and the overall health of their blood and brain, said Cheng Ching-Yu, head of SERI’s Ocular Epidemiology Research Group and Data Science Research Platform. “The retina is a non-invasive window into one’s biological age and systemic health status, and can tell us many things about a person’s morbidity and mortality risks”.
A person’s biological age, as compared to their chronological age, can better capture the physiological changes associated with ageing. Therefore, it can be used to assess their general health status. Cheng also claims that biological age is of greater interest to researchers, especially given the global ageing population and rising incidences of chronic diseases.
WHY IT MATTERS
Both RetiKid and RetiAge offer a non-invasive approach to screening health conditions in clinics. Patients may find them more bearable, possibly leading to more uptakes of and compliance with preliminary health screening.
Both tools can also be integrated with the Singapore Eye Lesion Analyser Plus, a retinal image-based system developed by SERI, which is currently being used by polyclinics across Singapore to screen for diabetic eye diseases, glaucoma and age-related macular degeneration.
For RetiKid, the screening process is automated, which allows for effective mass screening of at-risk patients. “Tools like RetiKid have [the] potential to be widely used in primary care to improve the current rate of CKD screening,” said Dr Cynthia Lim, a key investigator of the RetiKid project.
Additionally, RetiKid has the potential to be connected with smartphones, enabling point-of-care diagnosis. “Timely detection of CKD allows clinicians and patients the opportunity to intervene early and slow down [its] progression,” Dr Lim said.
Currently in its validation phase, RetiKid is going to be used in a community outreach programme by SERI in partnership with the National Kidney Foundation. The programme, which runs from February 2022 to January 2024, will involve about 1,200 participants with a high risk of developing CKD.
Meanwhile, researchers are currently working on refining the algorithms in RetiAge to optimise its prediction performance in the local population. They are also looking if it can also be used to predict other age-related diseases.
THE LARGER TREND
The Singapore National Eye Centre claims that the Singaporean population is “uniquely vulnerable” to eye diseases and disorders at every life stage. Myopia or nearsightedness affects one in two children by age 12. Among working adults, diabetic retinopathy is a major cause of vision loss. Moreover, the risk of blindness increases 15 times for Singaporeans aged 50 and above.
Among recent innovations in eye health in Singapore is an AI-based glaucoma screening method developed by a team of scientists and clinicians from the Nanyang Technological University and the Tan Tock Seng Hospital.
To help broaden the access to eye care in the country, Johnson & Johnson Vision has crafted a three-year roadmap to develop an integrated eye health ecosystem, focusing on data and digitisation initiatives. Some key projects it plans to implement include a community eye health e-referral network, an AI-powered eye care service, and telehealth.