This project aims to develop and validate an AI-based model for retinal age prediction using fundus photographs, and to examine the association between the retinal age gap (RAG) (the difference between predicted retinal age and chronological age) and systemic disease outcomes.
Research Questions:
Can retinal age be reliably predicted from fundus images using AI?
Can the retinal age gap serve as a non-invasive biomarker for risk stratification in population health?
Objectives:
To build and validate a deep learning model that predicts retinal age from retinal images.
To evaluate the correlation of the retinal age gap with systemic disease outcomes available in UK Biobank.
To assess the potential of RAG as a biomarker for identifying high-risk subgroups and informing preventive strategies.
Scientific Rationale:
The retina provides a unique, non-invasive view of microvasculature and neural tissue, making it an attractive biomarker of systemic health. Previous studies suggest that discrepancies between retinal age and chronological age are associated with increased morbidity and mortality. However, the clinical utility of this measure remains underexplored. By leveraging the scale and depth of UK Biobank data, alongside external datasets (e.g., ODIR), this study will validate AI-driven retinal age prediction and investigate its utility in systemic disease risk assessment.
This research has the potential to contribute to early detection and personalized preventive medicine by establishing the retinal age gap as a scalable, non-invasive biomarker.