AI spots your age—and fertility risks—just by scanning your eyes

New AI Accurately Estimates Age by Analyzing Retinal Images

Advances in generative AI have extended beyond text and image creation to enhanced computer vision capabilities. Modern AI models can interpret image data with impressive accuracy, enabling applications from photo editing to detailed image analysis. One emerging area involves medical imaging, where AI is beginning to assist in diagnostic and predictive tasks.

A recent study from China demonstrates how AI can determine a person’s age by analyzing retinal fundus images—photos of the back of the eye. This approach goes beyond estimating chronological age, as the AI also assesses the "retinal age gap," which reflects differences between the eye’s biological age and the person’s actual age.

Understanding Retinal Age and Its Medical Significance

Retinal fundus imaging reveals microvascular structures that correlate with systemic aging. The AI model used in this study, called Frozen and Learning Ensemble Crossover (FLEX), was trained on over 20,000 eye images from more than 10,000 adults. This extensive dataset allowed FLEX to learn patterns that indicate aging in retinal tissue.

Importantly, the researchers included a focus on pre-menopausal women, analyzing over 2,500 images to explore potential links between retinal aging and reproductive health.

Key Findings on Retinal Age and Reproductive Health

• The AI estimates a retinal age for each individual based on fundus photos.
• A positive retinal age gap—when the retina appears older than the actual age—may indicate accelerated aging of other organs.
• Among women aged 40 to 50, each additional retinal year increased the risk of lower anti-Müllerian hormone (AMH) levels, a marker for ovarian reserve, by 12% to 20% depending on the age bracket.
• An increased retinal age gap was also associated with a higher risk (36% per retinal year) of menopause before age 45.
• Women with more childbirths at younger ages showed lower AMH levels compared to average.

This suggests that retinal scans could serve as a non-invasive screening tool to assess reproductive aging and fertility potential. It offers a practical method for early intervention or planning, such as fertility preservation or hormonal therapies to manage menopause symptoms.

Potential Applications and Future Directions

While still in early stages, this research highlights the promise of AI in medical imaging beyond traditional diagnostics. A simple retinal scan could eventually help:

• Identify women at risk of early menopause.
• Guide decisions on fertility preservation in the late 20s or early 30s.
• Support personalized reproductive health care based on biological aging indicators.

Further studies are necessary to validate these findings and explore the FLEX model’s usefulness in detecting other age-related health conditions through retinal analysis.

For professionals interested in AI applications in healthcare and imaging, expanding expertise in AI-driven medical tools can be valuable. Resources like Complete AI Training offer relevant courses to deepen knowledge in this area.

The full study detailing these findings is published in Nature.