Related Experiment Videos
Clinical Positioning and Implementation of a Deep-Learning Retinal Biomarker (Reti-CVD) for Cardiovascular Risk
Junseung Rho1, Sung-Goo Kang1, Se-Hong Kim1
1Department of Family Medicine, St. Vincent's Hospital, College of Medicine, The Catholic University of Korea, Seoul 16247, Republic of Korea.
Journal of Clinical Medicine
|July 28, 2026
Summary
Retinal imaging AI (Reti-CVD) shows promise for cardiovascular disease (CVD) risk stratification, especially in borderline cases. However, more outcome-based studies are needed to confirm its clinical utility and guide integration.
Area of Science:
- Ophthalmology
- Cardiology
- Artificial Intelligence
Background:
- Cardiovascular disease (CVD) prevention relies on accurate risk stratification, but standard tools have limitations, particularly for borderline-risk individuals.
- Retinal imaging offers a non-invasive window into systemic microvasculature, with deep-learning oculomics potentially providing complementary CVD risk information.
- Reti-CVD, a retinal AI tool, generates a three-tier classification from retinal photographs and is among the more validated retinal AI tools for CVD risk assessment.
Purpose of the Study:
- To evaluate the clinical positioning and implementation of Reti-CVD as an exemplar of retinal AI in CVD risk stratification.
- To compare Reti-CVD's performance with established CVD risk scores and subclinical atherosclerosis markers.
- To consider the practical aspects of Reti-CVD implementation, including regulation and equity.
Main Methods:
- A narrative review of existing evidence on Reti-CVD, organized by cohort studies.
- Comparison of Reti-CVD's discriminative ability (Harrell C-index) with traditional risk scores.
- Analysis of Reti-CVD's reclassification improvement, particularly in borderline-risk populations.
Main Results:
- Reti-CVD demonstrated a discrimination of approximately 0.75 (Harrell C-index) with modest reclassification improvement, notably in borderline-risk groups.
- The tool has received marketing authorization in Korea (MFDS) and CE certification (EU MDR) as the commercial product DrNoon, but lacks US FDA authorization.
- Evidence primarily stems from a single research group and algorithm; no randomized or outcome-based trials confirm improved clinical outcomes with Reti-CVD-guided care.
Conclusions:
- Reti-CVD shows potential as a non-invasive risk enhancer for borderline/intermediate CVD risk reclassification.
- Current evidence is observational and hypothesis-generating, indicating a need for independent validation and intervention trials.
- Broad integration of Reti-CVD requires further evidence on clinical utility, cost-effectiveness, reimbursement, and demonstrated improvement in patient outcomes.