The clinical question
People with chronic kidney disease have heterogeneous cardiovascular risk. Retinal images offer a noninvasive view of the microvasculature.
The approach
The CARE-CKD study used deep learning to segment retinal vessels and machine learning to evaluate vascular features alongside clinical markers.
What the study adds
Retinal features added information for cardiovascular risk assessment in the CRIC cohort.
The next question
How well do these features generalize to other populations and prospective clinical settings?
Explore the source publication for study design, estimates, limitations, and full author contributions.