Researchers at the University of Dundee have used AI to detect if people with diabetes are at risk of developing chronic kidney disease years before symptoms emerge or current tests are able to detect problems.
This innovative new approach could allow for earlier intervention and treatment of diabetes-related chronic kidney disease. Diabetes is one of the most common causes of CKD, responsible for one in five people who need dialysis or a transplant.
Diabetes is a serious condition where glucose builds up in a person’s blood because it can’t get into their cells. Without treatment, high glucose levels in the blood can cause serious damage to the eyes, heart, feet and kidneys. There are two types of diabetes and together they affect around 5 million people in the UK.
In the UK, people living with diabetes over the age of 12 are regularly invited to have their eyes screened for signs of damage. A photo is taken of the retina, at the back of the eye. The researchers used nearly 1 million of these photos from almost 100,000 people with type 2 diabetes in Scotland for their research.
Linking the photos with existing data on kidney health, the researchers trained the AI to distinguish between images of people with kidney disease and those without. They also checked if the tool worked with data from almost 30,000 other people with type 2 diabetes.
The AI tool detected people with existing kidney disease with 86% accuracy. In people without kidney disease, it was able to predict who would go on to develop it in the next five years with 78% accuracy.
Significantly, the AI outperformed traditional kidney function tests, detecting kidney disease risk in people where the standard tests gave no warning.
By spotting who is at risk of developing kidney disease years before symptoms or current tests, the new tool could allow for earlier interventions in the future, helping millions of people avoid the devastating effects of kidney failure.
Using AI in this way is an innovative and novel approach and it is encouraging to see research indicating that kidney disease risk could be identified via routine diabetic eye screening. People can lose 90% of kidney function without any symptoms and, even if they do emerge, our research found that when presented with a list of kidney disease symptoms, 1 in 4 people were unable to identify these as signs of kidney disease. More accurate ways of identifying who is at greatest risk of kidney disease and spotting problems early means people can access treatment and support that could delay progression to more serious disease and potential kidney failure. This includes help to stop smoking and become more physically active.Fiona Loud, Policy Director at Kidney Care UK