Retinal Age as Disease Biomarker

May 27, 2026

The Ophthalmologist

Overview

A study by Ninomiya et al. demonstrates that a single fundus photograph can accurately estimate biological aging, termed 'retinal age,' with implications for systemic disease screening. The developed AI model shows promise in linking retinal age gaps to conditions such as diabetes and cardiac disease.

Background

The concept of retinal age is emerging within oculomics, aiming to quantify biological aging and disease risk through retinal imaging. Accurate assessment of retinal age could enhance disease screening and preventive care, particularly for systemic conditions. However, challenges in generalizability and clinical utility remain critical considerations.

Data Highlights

Validation Type	Mean Absolute Error (Years)
Internal	2.78
External (Hospital Cohort)	3.39
External (Larger Cohort)	Performance Declined

Key Findings

The AI model was trained on over 50,000 fundus images from more than 27,000 healthy individuals.
Retinal age gaps were significantly higher in patients with diabetes, cardiac disease, or a history of stroke.
The model achieved a mean absolute error of 2.78 years in internal validation.
External validation showed a mean absolute error of 3.39 years in a hospital cohort.
Attention was focused on the optic disc, macula, and major vascular arcades in predictions.

Clinical Implications

Integrating AI-based retinal age assessments into routine clinical workflows could enhance the identification of patients at risk for systemic diseases. This approach may facilitate early interventions without the need for additional tests.

Conclusion

The study supports the potential of retinal imaging as a non-invasive biomarker for systemic aging, suggesting a shift in its role from an ophthalmic tool to a broader preventive medicine platform.