AI‐based software for cardiovascular disease assessment

Abstract The healthcare burden of cardiovascular disease remains a priority. Understanding their underlying mechanisms and improving identification of people with higher systemic vascular risk profile with non‐invasive examinations seem to be crucial. In Ophthalmology, retinal imaging is simple, non‐invasive and it can provide in vivo insight of the retinal microcirculation. For more than 10 years, research teams have been working on software developed to allow automatic analysis of the retinal vascular network from different imaging technique (fundus photographs, oct‐angiography or adaptive optics…) and to provide a description of the geometric characteristics of its arterial and venous components. At the same time, research teams have explored the pathophysiological association between macro vascular and micro vascular alterations. Thus, retinal microcirculation could be considered as a witness of the systemic cardiovascular status. New approach with retinal images dataset and artificial intelligence algorithms recently strengthened the interest of retinal microvascular biomarker. Deep learning and machine learning algorithms provided very thorough description of the retinal vascular network. Moreover, they were able to assess strong associations between retinal features and calcium heart score or biological age for example. Thus, the interest of retinal biomarker in terms of screening, monitoring or prediction of cardiovascular disease is very promising.