Chronic kidney disease (CKD) can be accurately predicted even before symptoms appear using an artificial intelligence (AI) algorithm combined with a non-invasive retinal eye examination, according to a study just published in Nature Partner Journal's Digital Medicine by Mediwhale, Inc. and Yonsei University College of Medicine.
CKD is a leading cause of cardiovascular disease and related death, and its prevalence has grown rapidly due to aging populations and the increase in hypertension and diabetes patients. Symptoms do not materialize until the disease reaches advanced stages, and once it occurs, it is irreversible, making early detection and prevention of risk factors crucial.
Studies have shown that the current standard tests in CKD screening, the estimated Glomerular Filtration Rate (eGFR) conducted using a blood test or urine protein test, have shown limitations in accurately and early detecting kidney disease, especially in people with preserved kidney function.
In this study, Reti-CKD, a predictive risk score for CKD, was developed from a deep learning algorithm using retinal photographs. The rationale behind this approach lies in the shared characteristics between the retina and kidney, specifically their micro blood vessels. By leveraging the information obtained from the retinal blood vessels, Reti-CKD enables the prediction of kidney disease risk. Notably, Reti-CKD offers a convenient and cost-effective solution as it can be readily administered in primary care using a fundus camera within a few minutes.
To develop the deep learning model, retina data from over 79,000 Koreans' health checkups were used, and another 35,000 individuals from the UK Biobank and Korean Diabetic Cohort were employed in model validation.
Based on these data sets, Reti-CKD showed superior prediction of future CKD risk compared to the conventional eGFR or urine test. Persons identified as high-risk can take preemptive measures that will eliminate or reduce the risk of progression to kidney deterioration.
Also, Reti-CKD clearly stratified the risk of CKD into four classifications, ranging from low-risk to high-risk groups, and accurately predicted the risk regardless of the presence of underlying diseases such as hypertension or diabetes.
Dr. Tyler Hyungtaek Rim, Chief Medical Officer (CMO) of Mediwhale, said, "The significance of Reti-CKD is that it can prevent CKD by identifying high-risk groups and taking preemptive measures before kidney function declines. We anticipate that proactive risk management, diligent monitoring, and timely treatment will greatly benefit many people around the world."
Dr. Young Su Joo, professor in the Department of Nephrology at Yonsei University College of Medicine, said, "We aim to further extend our research in predicting kidney function deterioration among CKD patients. Our goal is to develop a prediction model that can effectively pre-screen the high-risk group for end-stage chronic kidney disease and evaluate its utility."