Нормативные характеристики сосудов сетчатки с учетом этнической принадлежности на основе данных UK Biobank с использованием глубокого обучения

Ethnicity-Stratified Normative Retinal Vascular Features from the UK Biobank Using Deep Learning.

The retina provides a unique, noninvasive window into systemic vascular health.^1,2 Artificial intelligence–based retinal analysis predicts cardiovascular disease with performance comparable to traditional biomarkers.^3,4 Clinical implementation requires robust normative data to generate individualized metrics such as the retinal age gap, which associates each year of deviation from expected values with a 2% increase in mortality risk and a 3% increase in cardiovascular disease risk.⁵ By identifying disease-specific biomarkers through artificial intelligence algorithms and defining their normative ranges, clinicians can identify high-risk patients and detect systemic diseases earlier.^6,7 In this study, we address this need by establishing ethnicity-stratified normative ranges for retinal vascular features in 6843 UK Biobank participants.