Article Synopsis
- A new AI deep learning system called DeepDKD was developed to improve screening for diabetic kidney disease (DKD) and differentiate between isolated diabetic nephropathy and non-diabetic kidney disease (NDKD) using retinal fundus images.
- The system was trained on a massive dataset of over 734,000 retinal images and validated across multiple populations including participants from China, Singapore, and the UK to ensure accuracy.
- Results showed DeepDKD had a strong performance, with an area under the curve (AUC) of 0.842 for DKD detection and 0.906 for differentiating nephropathy types, indicating its potential as an effective screening tool in diabetes care.
Video Abstracts
Category Ranking
98%
Total Visits
921
Avg Visit Duration
2 minutes
Citations
20
Article Abstract
Background: Improving the accessibility of screening diabetic kidney disease (DKD) and differentiating isolated diabetic nephropathy from non-diabetic kidney disease (NDKD) are two major challenges in the field of diabetes care. We aimed to develop and validate an artificial intelligence (AI) deep learning system to detect DKD and isolated diabetic nephropathy from retinal fundus images.
Methods: In this population-based study, we developed a retinal image-based AI-deep learning system, DeepDKD, pretrained using 734 084 retinal fundus images. First, for DKD detection, we used 486 312 retinal images from 121 578 participants in the Shanghai Integrated Diabetes Prevention and Care System for development and internal validation, and ten multi-ethnic datasets from China, Singapore, Malaysia, Australia, and the UK (65 406 participants) for external validation. Second, to differentiate isolated diabetic nephropathy from NDKD, we used 1068 retinal images from 267 participants for development and internal validation, and three multi-ethnic datasets from China, Malaysia, and the UK (244 participants) for external validation. Finally, we conducted two proof-of-concept studies: a prospective real-world study with 3 months' follow-up to evaluate the effectiveness of DeepDKD in screening DKD; and a longitudinal analysis of the effectiveness of DeepDKD in differentiating isolated diabetic nephropathy from NDKD on renal function changes with 4·6 years' follow-up.
Findings: For detecting DKD, DeepDKD achieved an area under the receiver operating characteristic curve (AUC) of 0·842 (95% CI 0·838-0·846) on the internal validation dataset and AUCs of 0·791-0·826 across external validation datasets. For differentiating isolated diabetic nephropathy from NDKD, DeepDKD achieved an AUC of 0·906 (0·825-0·966) on the internal validation dataset and AUCs of 0·733-0·844 across external validation datasets. In the prospective study, compared with the metadata model, DeepDKD could detect DKD with higher sensitivity (89·8% vs 66·3%, p<0·0001). In the longitudinal study, participants with isolated diabetic nephropathy and participants with NDKD identified by DeepDKD had a significant difference in renal function outcomes (proportion of estimated glomerular filtration rate decline: 27·45% vs 52·56%, p=0·0010).
Interpretation: Among diverse multi-ethnic populations with diabetes, a retinal image-based AI-deep learning system showed its potential for detecting DKD and differentiating isolated diabetic nephropathy from NDKD in clinical practice.
Funding: National Key R & D Program of China, National Natural Science Foundation of China, Beijing Natural Science Foundation, Shanghai Municipal Key Clinical Specialty, Shanghai Research Centre for Endocrine and Metabolic Diseases, Innovative research team of high-level local universities in Shanghai, Noncommunicable Chronic Diseases-National Science and Technology Major Project, Clinical Special Program of Shanghai Municipal Health Commission, and the three-year action plan to strengthen the construction of public health system in Shanghai.
Download full-text PDF |
Source |
|---|---|
| http://dx.doi.org/10.1016/j.landig.2025.02.008 | DOI Listing |
Publication Analysis
Top Keywords
Similar Publications
Influence of dietary components on the gut microbiota of middle-aged adults: the gut-Mediterranean connection.
BMC Microbiol
September 2025
Faculty of Kinesiology, University of Calgary, Calgary, AB, Canada.
Background: A plant-focused, healthy dietary pattern, such as the Mediterranean diet enriched with dietary fiber, polyphenols, and polyunsaturated fats, is well known to positively influence the gut microbiota. Conversely, a processed diet high in saturated fats and sugars negatively impacts gut diversity, potentially leading to weight gain, insulin resistance, and chronic, low-grade inflammation. Despite this understanding, the mechanisms by which the Mediterranean diet impacts the gut microbiota and its associated health benefits remain unclear.
View Article and Find Full Text PDFPortal vein-enriched metabolites as intermediate regulators of the gut microbiome in insulin resistance.
Cell Metab
August 2025
Section of Integrative Physiology and Metabolism, Joslin Diabetes Center, Harvard Medical School, Boston, MA, USA. Electronic address:
Diet and obesity contribute to insulin resistance and type 2 diabetes, in part via the gut microbiome. To explore the role of gut-derived metabolites in this process, we assessed portal/peripheral blood metabolites in mice with different risks of obesity/diabetes, challenged with a high-fat diet (HFD) + antibiotics. In diabetes/obesity-prone C57BL/6J mice, 111 metabolites were portally enriched and 74 were peripherally enriched, many of which differed in metabolic-syndrome-resistant 129S1/129S6 mice.
View Article and Find Full Text PDFSocial isolation promotes hyperglycemia through sympathetic activation of inguinal white adipose tissue.
Biochem Biophys Res Commun
September 2025
Yangzhi Rehabilitation Hospital (Shanghai Sunshine Rehabilitation Center), Tongji University School of Medicine, Shanghai, PR China. Electronic address:
Epidemiological studies have reported that social isolation increases the risk of diabetes, but the underlying neural mechanism remains unclear. Using a long-term single-housed (SH) mouse model of social isolation, SH mice not only exhibited disrupted glucose homeostasis, evidenced by elevated fasting glucose, impaired glucose tolerance, and reduced insulin sensitivity, but also showed hypertrophic adipocytes and altered lipid metabolism. To elucidate the neural mechanisms underlying these metabolic disturbances, retrograde trans-synaptic tracing revealed the paraventricular nucleus (PVN) and locus coeruleus (LC) as the most PRV-labeled brain regions, suggesting their potential roles in social isolation-induced hyperglycemia.
View Article and Find Full Text PDFDiverse social environments and 10-year all-cause mortality in a cohort of historically excluded racial/ethnic older adults: A latent class analysis of social integration.
Ann Epidemiol
September 2025
Department of Health Promotion Sciences, The University of Arizona.
Purpose: The association between social integration and mortality in older adults from historically excluded groups may not align with the patterns observed in predominately white samples. We modeled latent groups of social integration and their association with 10-year all-cause mortality in a cohort of older adults from historically excluded racial/ethnic groups.
Methods: In a sub-sample of a national cohort study of older adults, we used latent class analysis to model social integration using ten item indicators spanning couple status, network characteristics, and neighborhood and community connections.
Safety and resource utilisation efficiency of semi-skeletonised versus skeletonised left internal mammary artery harvesting techniques: The BANGABANDHU study.
PLoS One
September 2025
Department of Cardiac Surgery, Bangabandhu Sheikh Mujib Medical University, Dhaka, Bangladesh.
Background: The ideal harvesting techniques of the left internal mammary artery (LIMA) for coronary artery bypass graft (CABG) are elusive. We assessed the safety and resource utilisation efficiency of semi-skeletonised LIMA harvesting techniques, focusing on length, harvesting time, and the number of Ligaclips used compared to skeletonised techniques within a single surgeon's practice.
Methods: The BANGABANDHU (Bangladeshi Atherosclerosis Biobank AND Hub) study was an ambispective observational cohort that evaluated age- and sex-matched 2209 adult Bangladeshi isolated CABG population from 1st January 2015 to 31 January 2025.