Real-World Prospective Validation and Economic Evaluation of Deep Learning– Based Diabetic Retinopathy Detection From Fundus Photographs: A Systematic Review and Meta-analysis
Authors' conclusion
DL-DR systems using FPs had high discriminative performance in prospective real-world settings and hold promise to improve cost-effectiveness, especially in high-income countries.
Study context
- Population
- Prospective real-world validation and economic studies of deep learning (DL)-based diabetic retinopathy (DR) detection systems using fundus photographs47 studies
- Outcome
- Pooled diagnostic performance (AUROC) of DL-DR systems
Methodological detailsprospective validation studies, economic evaluation, Five databases searched until 13 August 2025
- Prospective validation studies
- Economic evaluation
- Five databases searched until 13 August 2025
Pooled result
As reported in the abstract
Pooled AUROC by DR severity
AUROC, vision-threatening DR
0.974
AUROC, any DR
0.965
AUROC, referable DR
0.959
Further pooled results
Factors affecting referable DR (RDR) detection performance
- Study region, clinical pathway, mydriasis, image quality control, sample size, grading criteria, reference standard and model architecture significantly affected performance in RDR detection
Cost-effectiveness (15 studies)
- DL-based DR screening cost-effective in high-income countries; mixed in middle-income countries, depending on compliance rates, glycemic control and initial costs
Original publication
Real-World Prospective Validation and Economic Evaluation of Deep Learning– Based Diabetic Retinopathy Detection From Fundus Photographs: A Systematic Review and Meta-analysis. Diabetes Care. Published 19 November 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →