Abstract 16742: Deep-Learning-Derived Retinal Cardiovascular Risk Predictor (Reti-CVD) and 14 Cardiovascular Conditions in UK Biobank

Introduction: The advent of sophisticated deep learning algorithms has now made it possible to predict the risk of cardiovascular diseases (CVDs) using retinal images. We had previously developed a retina-based deep learning model, Reti-CVD, trained on coronary artery calcium (CAC) data, which successfully predicted future CVD incidents in a longitudinal study. Hypothesis: This study aims to investigate the cross-sectional association between Reti-CVD and 13 distinct CVDs, alongside arterial hypertension. Methods: Our cross-sectional analysis involved 45,980 participants from the UK Biobank at baseline. To discern the differential cardiovascular risk associated with Reti-CVD, we studied a wide array of CVDs. These included cerebrovascular diseases, aneurysms, thrombo-embolic diseases, other CVDs (coronary artery disease, aortic valve stenosis, atrial fibrillation, heart failure, and peripheral vascular disease), and arterial hypertension. We defined CVD outcomes based on the international classification of disease codes. We used logistic regression, adjusted for hypertension, diabetes, dyslipidemia, and smoking, to estimate the correlations between Reti-CVD and the defined CVD outcomes. Results: In the cross-sectional study, after adjusting for CVD risk factors, we found the highest tertile of Reti-CVD scores to be significantly associated with 11 outcomes in comparison to the first tertile (adjusted Odds Ratio [OR], 95% Confidence Interval [CI]). These include: Coronary artery disease (OR=10.37, 95% CI, 7.58-14.18), peripheral vascular disease (9.65, 2.94-31.64), atrial fibrillation (9.36, 6.51-13.45), aortic valve stenosis (8.13, 1.87-35.35), heart failure (7.33, 3.64-14.77), ischemic stroke (4.70, 2.30-9.59), transient ischemic attack (4.17, 2.07-8.40), arterial hypertension (3.17, 2.78-3.61), pulmonary embolism (3.00, 1.86-4.84), deep vein thrombosis (2.54, 1.70-3.80), and cerebrovascular diseases (2.36, 1.63-3.42). Notably, we found suggestive evidence of an inverse association of Reti-CVD tertiles with subarachnoid haemorrhage. Conclusions: This study demonstrates that higher Reti-CVD scores correlate with an elevated risk across a broad range of cardiovascular conditions.