Multi-omic analysis of deep learning-derived phenotypes links ophthalmic imaging to cardiovascular and neurological traits.

The eye is a recognized source of biomarkers for cardiovascular and neurodegenerative disease risk. Here we characterize the breadth of these associations and identify biological axes that may mediate them. Using UK Biobank data, we developed a multi-omic analysis pipeline integrating physiological, radiomic, metabolomic and genomic information. We trained retinal adversarial autoencoders to represent optical coherence tomography images and color fundus photographs as 256-dimensional embeddings. Retinal adversarial autoencoder-derived embeddings were associated with a range of cardiovascular and neurodegenerative diseases, including ischemic heart disease, cerebrovascular disease, Parkinson's disease and dementia. Examining associations across diverse omics datasets, we provide evidence linking ophthalmic imaging features to neurological and cardiovascular anatomy and function, lipid metabolism and gene sets associated with neurodegenerative pathology. Collectively, our findings show that ophthalmic features reflect complex, multisystem biological processes and reinforce the role of the eye as a composite indicator of systemic health.