Abstract: SA-PO0019
Retinal Biomarkers Reveal Distinct Patterns of Kidney Risk Across Metabolic Health Phenotypes
Session Information
- Intelligent Imaging and Omics: Phenotyping and Risk Stratification
November 08, 2025 | Location: Exhibit Hall, Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Artificial Intelligence, Digital Health, and Data Science
- 300 Artificial Intelligence, Digital Health, and Data Science
Authors
- Thakur, Sahil, Mediwhale Inc, Seoul, Korea (the Republic of)
- Rukmini, Annadata V., Mediwhale Inc, Seoul, Korea (the Republic of)
- Park, Junseok, Mediwhale Inc, Seoul, Korea (the Republic of)
- Nam, Dongjin, Mediwhale Inc, Seoul, Korea (the Republic of)
- Cho, Jungkyung, Mediwhale Inc, Seoul, Korea (the Republic of)
- Park, Tae Hyun, Mediwhale Inc, Seoul, Korea (the Republic of)
- Seo, Jaewon, Mediwhale Inc, Seoul, Korea (the Republic of)
- Nusinovici, Simon, Mediwhale Inc, Seoul, Korea (the Republic of)
- Rim, Tyler Hyungtaek, Mediwhale Inc, Seoul, Korea (the Republic of)
Background
Retinal imaging offers a non-invasive window into microvascular health, potentially reflecting kidney disease risk. This study evaluated deep learning (DL)-derived retinal biomarkers across metabolic health phenotypes to examine associations with kidney function.
Methods
We analyzed retinal images from UK Biobank participants (N=59,493), categorizing them into six phenotypes: metabolically healthy normal weight (MHN), overweight (MHOW), obese (MHO); and metabolically unhealthy normal (MUN), overweight (MUOW), obese (MUO). Metabolic health was defined as blood pressure <130 mmHg without antihypertensives, appropriate waist-to-hip ratio, and absence of diabetes. DL-derived retinal scores included Reti-CKD (predicting CKD), Reti-ACR (albumin-to-creatinine ratio), and Reti-eGFR (estimated glomerular filtration rate).
Results
All retinal biomarkers showed clear differentiation between metabolically healthy and unhealthy groups. Unhealthy phenotypes had significantly higher Reti-CKD, Reti-ACR, and Reti-eGFR scores, indicating worse predicted kidney function. Within both health categories, biomarker scores increased stepwise from normal weight to obesity. MUO individuals had the highest risk profiles across all scores.
Conclusion
DL-derived retinal biomarkers effectively reveal kidney risk stratification across metabolic phenotypes. Metabolically unhealthy obesity (MUO) showed the strongest link with adverse kidney indicators. These findings support retinal imaging as a valuable, non-invasive tool for early kidney risk detection, particularly in metabolically at-risk populations.
Boxplot of Reti-CKD scores across six metabolic phenotypes: MHN, MHOW, MHO, MUN, MUOW, MUO. Asterisks indicate statistically significant differences between groups.