AI Model Uses Eye Imaging to Identify Risk of Major Systemic Diseases

Early detection of systemic disease risk remains a persistent challenge in population health screening. Cardiometabolic conditions such as diabetes, heart disease, and stroke often progress without symptoms until complications arise. Clinicians need noninvasive tools that fit within routine care and can triage patients for targeted evaluation. Researchers have now developed an artificial intelligence (AI) approach that estimates retinal age from a single fundus photograph and links deviations from chronological age to elevated risk for these major diseases.

A research group at the Tohoku University Graduate School of Medicine (Sendai, Japan) developed an AI model that derives an individual’s “retinal age” from one color fundus image. The model analyzes retinal features that reflect biological aging. It was trained on 50,595 quality-controlled fundus images from disease‑free adults and internally validated on 7,288 additional images.

During training, the algorithm incorporated hemoglobin A1c (HbA1c) to help the network capture age‑related retinal patterns more robustly. The system achieved an average absolute error of about three years when estimating chronological age. If deployed clinically, no blood test would be required because the output is generated directly from the retinal photograph.

Investigators then examined the “retinal age gap,” defined as the difference between the AI‑predicted retinal age and a person’s actual age. After matching participants by age and sex, the gap was significantly larger in individuals with diabetes, heart disease, or a history of stroke, indicating that their retinas appeared older than expected. The analyses were primarily cross‑sectional, indicating correlation rather than causation, and the team noted the need for prospective longitudinal studies.

External validation of the study was conducted with support from University College London, a strategic partner of Tohoku University. The findings were published in Communications Medicine on April 8, 2026. The tool is designed to serve as a screening aid that could flag patients for further assessment or prevention strategies at a clinician’s discretion.

“Fundus images are non-invasive photos of the eye taken as part of regular health check-ups - so no additional work is needed, Our model would be a nearly frictionless addition to a clinician's typical workflow,” said Toru Nakazawa, Professor at the Tohoku University Graduate School of Medicine.

“We are already planning a study that follows a cohort of over 10,000 individuals with continuous 3-year follow-up to examine whether retinal age-related signals are associated with the future development of cardiovascular and other systemic diseases,” said Prof. Nakazawa.

Related Links
Tohoku University