Systematic review: meta-analysis of 47 prospective studies found deep learning detection of diabetic retinopathy from fundus photographs performed well (AUROC 0.959 for referable disease); 15 economic studies found it cost-effective in high-income countries, mixed elsewhere.
Meta-analysisn = 47 studies
Ran et al. Real-World Prospective Validation and Economic Evaluation of Deep Learning- Based Diabetic Retinopathy Detection From Fundus Photographs: A Systematic Review and Meta-analysis