A noninvasive model for chronic kidney disease screening and common pathological type identification from retinal images

作者：Qianni Wu, Jianbo Li, Lanqin Zhao, Dong Liu, Jingyi Wen, Yunuo Wang, Yiqin Wang, Naya Huang, Lanping Jiang, Qinghua Liu, Hanming Lin, Hanming Lin, Pengxia Wan, Shicong Yang, Wenfang Chen, Wenfang Chen, Hongjian Ye, Mohammed Haji Rashid Hassan, Ahmed Hassan Nur, Zefang Dai, Jianwen Yu, Shanshan Zhou, Jianwen Yu, Weixing Zhang, Wenben Chen, Wenben Chen, Ruiyang Li, Wai Cheng Iao, Juanjuan Feng, Yan Wang, Hua Hong, Peihong Yin, Qing Ye, Chao Xie, Min Zhu, Xiaoyi Liu, Yaozhong Kong, Jie Wang, Rongguo Fu, Xiao Yu, Guoguang Chen, Rongguo Fu, Yuhe Ke, J Ling, Charumathi Sabanayagam, Daniel Shu Wei Ting, Kar Keung Cheng, Duoru Lin, Wei Chen, Wei Chen, Haotian Lin, Haotian Lin · 发表于：Nature Communications · 年份：2025 · DOI：10.1038/s41467-025-62273-0 · 被引用次数：10 · 研究领域：Retinal Imaging and Analysis、Retinal and Optic Conditions、Artificial Intelligence in Healthcare

Chronic kidney disease (CKD) is a global health challenge, but invasive renal biopsies, the gold standard for diagnosis and prognosis, are often clinically constrained. To address this, we developed the kidney intelligent diagnosis system (KIDS), a noninvasive model for renal biopsy prediction using 13,144 retinal images from 6773 participants. The KIDS achieves an area under the receiver operating characteristic curve (AUC) of 0.839–0.993 for CKD screening and accurately identifies the five most common pathological types (AUC: 0.790–0.932) in a multicenter and multi-ethnic validation, outperforming nephrologists by 26.98% in accuracy. Additionally, the KIDS further predicts disease progression based on pathological classification. Given its flexible strategy, the KIDS can be adapted to local conditions to provide a tailored tool for patients. This noninvasive model has the potential to improve CKD clinical management, particularly for those who are ineligible for biopsies. The diagnosis and prognosis of chronic kidney disease (CKD) relies on invasive renal biopsies, which can be clinically constrained. Here, the authors develop a non-invasive model using retinal images that can be used to screen for CKD and identify five common pathological types of CKD.