AI-Powered Protein Clocks Detect Accelerated Eye Aging Before Disease Strikes
Proteomic aging clocks combined with deep learning reveal premature biological aging in cataract, glaucoma, AMD, and diabetic retinopathy across 55,000 people.
Summary
Researchers combined high-throughput blood proteomics with AI-driven eye imaging to build biological aging clocks capable of detecting accelerated aging in four major age-related eye diseases. Using data from over 55,000 participants across multiple countries and ethnicities, they showed that proteomic aging scores outperform chronological age in predicting cataract, diabetic retinopathy, age-related macular degeneration, and glaucoma. The clocks also linked faster biological aging to structural damage in retinal nerve tissue and microvascular loss. A streamlined, cost-effective version of the clock retained strong predictive accuracy. The findings suggest proteomic aging acceleration is a scalable, cross-ethnic biomarker for tracking eye health and shared aging pathways underlying multiple ocular diseases simultaneously.
Detailed Summary
Age-related eye diseases are among the leading causes of preventable blindness worldwide, yet current diagnostic tools largely respond to damage already done. This study asks whether biological aging measured through blood proteins — combined with AI analysis of retinal images — can detect disease risk earlier and more precisely than a person's chronological age alone.
The researchers drew on data from over 55,000 participants across three large, ethnically diverse international cohorts. They used high-throughput proteomics to build machine-learning-based aging clocks, then validated these clocks against deep-learning-derived eye phenotypes extracted from advanced retinal imaging, including optical coherence tomography and angiography.
The proteomic clocks successfully identified premature biological aging in individuals with all four major age-related eye diseases: cataract, diabetic retinopathy, age-related macular degeneration (AMD), and glaucoma. Crucially, proteomic aging acceleration predicted these conditions beyond what chronological age alone could explain. The clocks performed consistently across sexes and ethnic groups. A simplified, lower-cost version of the proteomic clock preserved most of the predictive power, pointing toward real-world clinical utility.
By linking accelerated proteomic aging to both neuroretinal degeneration and microvascular rarefaction in the retina, the study reveals a coupled neural-vascular aging signature in the eye. This dual deterioration pattern — nerve loss alongside blood vessel dropout — suggests shared biological pathways drive multiple ocular diseases simultaneously rather than each condition arising independently.
The implications extend beyond ophthalmology. The retina is often called a window into systemic aging, and proteomic clocks that track eye aging may also reflect broader whole-body aging trajectories. Limitations include reliance on abstract-only data for this summary, cross-sectional design elements that limit causal inference, and the need for prospective validation of the streamlined clock in clinical settings.
Key Findings
- Proteomic aging clocks predicted cataract, AMD, glaucoma, and diabetic retinopathy beyond chronological age across 55,000+ participants.
- Accelerated proteomic aging linked to both retinal nerve fiber loss and microvascular rarefaction, indicating coupled neural-vascular decline.
- A simplified, low-cost version of the proteomic clock maintained strong predictive accuracy, supporting clinical scalability.
- Clock performance held consistently across sexes and multiple ethnic groups, suggesting broad applicability.
- Deep learning retinal imaging and blood proteomics together reveal shared aging pathways underlying multiple distinct eye diseases.
Methodology
The study used data from three large cross-national cohorts totaling over 55,000 transethnic participants. High-throughput plasma proteomics was paired with deep-learning pipelines analyzing optical coherence tomography and angiographic retinal images to derive structural and vascular biomarkers. Machine learning models were trained in discovery cohorts and externally validated, including in the Guangzhou Diabetic Eye Study and the HOPE study.
Study Limitations
This summary is based on the abstract only, as the full text was not accessible. The study design includes cross-sectional elements that limit causal inference between proteomic aging acceleration and disease onset. Prospective clinical validation of the streamlined proteomic clock in diverse real-world healthcare settings is still needed.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.
Enter your email to subscribe: