Skip to navigation Skip to main content Skip to footer

Approved Research

Predicting cardiovascular diseases from retina imaging and multimodal data

Principal Investigator: Dr Sayoko Moroi
Approved Research ID: 89579
Approval date: May 2nd 2023

Lay summary

Cardiovascular diseases are a major public health problem and the leading cause of death. The current standard-of-care for screening cardiovascular disease includes age, gender, smoking status, blood pressure, body mass index, glucose, and cholesterol levels. This screening is predictive for cardiovascular disease, but it does not make a personal impact on a large proportion of individuals as cardiovascular disease remains a leading cause of death. There is gap in knowledge to provide individual risk for cardiovascular disease to prompt life-style changes for better health. We propose that pictures or images of the retina blood vessels will inform us about the blood vessels in other parts of the body as the retina provides the literal window into systemic health. Using deep learning analytical tools to study the retina images, newer studies show promise to predict cardiovascular disease. The value of the retina images is direct evidence of the impact of the clinical risk factors on an individual's blood vessel. The retina image will be labeled and annotated with the individual's clinical risk factors data that are present in the UK Biobank database. The expected outcome from this annotated dataset will be individualized cardiovascular risk stratification models from an individual's retina image. The generalization of these models will be tested in other datasets representing other populations. If replication of such models in other datasets is demonstrated, then there are considerable opportunities for public health-based implementation science for physicians, nurses, dieticians, other related health professionals, patients, and health insurance systems. The advantage of adding retina imaging to the risk stratification is to show the individual the effect of his/her behaviors on cardiovascular health and risk for potential illness from heart disease. The expected significance of this retina image-based risk stratification model is to prompt the patient for life-style changes towards better health.