Vascular Retinomics Decodes Panvascular Progression: A Retinal Polyvascular Disease Risk Index

Purpose: To evaluate whether color fundus photograph-derived vascular retinomics predicts progression from single vascular disease to polyvascular disease. Methods: This prospective cohort study included UK Biobank participants with baseline single vascular disease and eligible color fundus photographs. Progression was defined as incident involvement of a previously unaffected coronary, cerebrovascular, or peripheral arterial bed. After LASSO-Cox and Boruta feature selection, eight survival-learning algorithms were compared. The best-performing model was refitted to derive the Retinal Polyvascular Disease Risk Index (RPVDRI). Performance was evaluated in the testing cohort and benchmarked against conventional clinical factors and SMART2. Results: Among 2592 participants, 334 progressed during a median follow-up of 11.39 years. A 25-feature vascular retinomic panel was retained, and a gradient boosting machine performed best. In the testing cohort, the RPVDRI showed a graded association with progression after multivariable adjustment (highest vs. lowest tertile: hazard ratio, 1.97; 95% confidence interval, 1.03-3.74; P = 0.040; P for trend = 0.029). The RPVDRI achieved 3-, 5-, and 10-year areas under the curve (AUCs) of 0.636, 0.716, and 0.628, respectively. Adding the RPVDRI to conventional clinical factors improved 5-year AUC by 0.113 (P = 0.002), whereas the corresponding improvement beyond SMART2 was 0.037 (P = 0.466). Shapley additive explanations linked predictions to multidomain features and co-attribution. Conclusions: The RPVDRI was independently associated with polyvascular disease progression and may complement clinical risk assessment in secondary prevention, particularly in scalable or opportunistic settings where comprehensive clinical information is limited. Translational Relevance: Fundus-derived vascular retinomics may support precision secondary prevention by noninvasively stratifying residual polyvascular progression risk.

Authors

Institutions

Publication Details

Journal
Translational Vision Science & Technology
Published
2026-11-16
DOI
https://doi.org/10.1167/tvst.15.11.15
Primary Topic
Retinal Imaging and Analysis
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Vascular Retinomics Decodes Panvascular Progression: A Retinal Polyvascular Disease Risk Index

Tengda Huang, Zhishen Peng, Zhenchao Du, Xiaohong Yang et al.
2 links
Translational Vision Science & Technology
Retinal Imaging and Analysis
article

Vascular Retinomics Decodes Panvascular Progression: A Retinal Polyvascular Disease Risk Index

Tengda Huang, Zhishen Peng, Zhenchao Du, Xiaohong Yang, Jianuo Han, Weijia Chen, Lei Jiang, Yao Ying, Jingyan Peng, Yan Wang, Honghua Yu, Danli Shi, Lei Liu, Yi Zhang, Heng Li, Dan Jouma A. Maman Lawali, Wenli Zhang, Yun Ren, Cong Li, Xue He, Sunjin Hwang
article en

Abstract

Purpose: To evaluate whether color fundus photograph-derived vascular retinomics predicts progression from single vascular disease to polyvascular disease. Methods: This prospective cohort study included UK Biobank participants with baseline single vascular disease and eligible color fundus photographs. Progression was defined as incident involvement of a previously unaffected coronary, cerebrovascular, or peripheral arterial bed. After LASSO-Cox and Boruta feature selection, eight survival-learning algorithms were compared. The best-performing model was refitted to derive the Retinal Polyvascular Disease Risk Index (RPVDRI). Performance was evaluated in the testing cohort and benchmarked against conventional clinical factors and SMART2. Results: Among 2592 participants, 334 progressed during a median follow-up of 11.39 years. A 25-feature vascular retinomic panel was retained, and a gradient boosting machine performed best. In the testing cohort, the RPVDRI showed a graded association with progression after multivariable adjustment (highest vs. lowest tertile: hazard ratio, 1.97; 95% confidence interval, 1.03-3.74; P = 0.040; P for trend = 0.029). The RPVDRI achieved 3-, 5-, and 10-year areas under the curve (AUCs) of 0.636, 0.716, and 0.628, respectively. Adding the RPVDRI to conventional clinical factors improved 5-year AUC by 0.113 (P = 0.002), whereas the corresponding improvement beyond SMART2 was 0.037 (P = 0.466). Shapley additive explanations linked predictions to multidomain features and co-attribution. Conclusions: The RPVDRI was independently associated with polyvascular disease progression and may complement clinical risk assessment in secondary prevention, particularly in scalable or opportunistic settings where comprehensive clinical information is limited. Translational Relevance: Fundus-derived vascular retinomics may support precision secondary prevention by noninvasively stratifying residual polyvascular progression risk.

Translational Vision Science & TechnologyVol. 15(11)
Hong Kong Polytechnic University (HK), Chinese University of Hong Kong (HK), Shantou University (CN), Shantou University Medical College (CN), Hanyang University Guri Hospital (KR), Hanyang University Seoul Hospital (KR), Guangdong Academy of Medical Sciences (CN), Guangdong Provincial People's Hospital (CN), Joint Shantou International Eye Center (CN), Hanyang University (KR), Southern Medical University (CN)
Good health and well-being
Retinal Imaging and Analysis
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon