identifeye HEALTH is an AI company developing a fully automated fundus camera and end-to-end diagnostic systems to scale retinal screening in primary care and low-resource settings. The retina provides a non-invasive window into microvascular and neurodegenerative changes linked to both ocular and systemic disease. This project will use UK Biobank retinal images, linked health records, and longitudinal outcomes to train and validate AI models for detection and severity staging of diabetic retinopathy (DR), age-related macular degeneration (AMD), and glaucoma, and for risk assessment of systemic disease (coronary artery disease, chronic kidney disease, hypertension, and Alzheimer’s-related cognitive decline proxies).
Research questions:
1) How accurately can AI perform healthy-vs-any triage, detect, and precisely stage DR/AMD/glaucoma, and how do single-disease models compare with unified multi-disease (foundation-style) models, including cross-disease pretraining/bootstrapping to improve robustness?
2) How accurately can AI detect and predict systemic outcomes (CAD, CKD, hypertension, and cognitive decline proxies) using retinal biomarkers and longitudinal data?
3) Can high-resolution explainability methods generate clinically useful maps that localize retinal features associated with disease severity and risk?
Objectives: develop and validate scalable deep learning models; quantify retinal imaging biomarkers; evaluate generalizability across devices, acquisition conditions, and demographic subgroups with calibration and fairness analyses; and deliver explainability outputs suitable for clinical review and potential feature detection.
Scientific rationale: retinal vasculature, optic nerve, and macular structures reflect systemic microvascular injury and neurodegeneration. UK Biobank’s scale and linkage enable rigorous validation and biomarker discovery to support equitable, clinically deployable screening.
