A noninvasive intelligent kidney disease diagnosis system based on retinal color photographs, the Kidney Intelligent Diagnosis System (KIDS), has been developed by the Intelligent Eye-Kidney Alliance (iEKA), according to the Zhongshan Ophthalmic Center (ZOC) at Sun Yat-sen University.
This cross-institutional alliance was led by Professor Lin Haotian's team from ZOC and Professor Chen Wei's team from the Department of Nephrology at the First Affiliated Hospital of Sun Yat-sen University (FAH-SYSU), in collaboration with multiple hospitals both domestically and internationally.
On July 29, the research findings were published online in Nature Communications, with one invention patent granted. This breakthrough provides technical support for kidney disease screening and precision management in underdeveloped regions and primary healthcare institutions while establishing a new paradigm for the coordinated management of eye and kidney diseases.
Research paper was published online in Nature Communications on July 29, 2025.
Chronic Kidney Disease (CKD) has emerged as a major global public health challenge due to its high prevalence, disability rate, and mortality. In May 2025, the World Health Organization formally designated kidney diseases as a priority non-communicable disease requiring worldwide attention.
Renal biopsy remains the gold standard for definitive CKD diagnosis and management. However, this invasive procedure faces significant limitations in underdeveloped regions and primary care settings due to stringent requirements for patient indications and operator expertise, coupled with inherent risks of complications.
To address this challenge, ophthalmologists from ZOC and nephrologists from FAH-SYSU established iEKA in July 2021.
The team collected 13,144 retinal images in preliminary research, developing a multimodal machine learning model to create KIDS. The system underwent multicenter validation across the Guangdong-Hong Kong-Macao Greater Bay Area, Xinjiang Uygur Autonomous Region, and Somalia, Africa.
Operating solely on retinal fundus inputs, KIDS performs three core functions: CKD screening, pathological diagnosis, and prognostic prediction, achieving comprehensive disease management across the entire clinical pathway.
KIDS assists physicians in screening, diagnosing, and predicting prognosis for CKD.
Furthermore, to address severe medical resource constraints and limited renal biopsy access in underdeveloped regions like Somalia, the team developed a streamlined variant of KIDS. This simplified model accurately predicts CKD pathological types using only minimal blood and urine test parameters. Validated across global real-world multicenter datasets, it achieved an AUC of 0.952 (AUC measures diagnostic accuracy; values closer to 1 indicate higher precision), demonstrating broad applicability in resource-limited settings.
Reporter | Luo Shuxin (intern), Chen Jinxia
Photo | Zhongshan Ophthalmic Center
Editor | Hu Nan, James, Shen He
