Diabetic Retinopathy Screening Approaches in Developing Countries: A Systematic Review and Meta-Analysis

Area of Science:

• Ophthalmology
• Medical Technology
• Public Health

Background:

• Diabetic retinopathy (DR) is a leading cause of vision loss, with increasing global prevalence.
• Effective screening is crucial, especially in developing nations with limited healthcare resources.
• Current screening methods face challenges in accessibility and specialized personnel.

Purpose of the Study:

• To evaluate the diagnostic accuracy of different diabetic retinopathy (DR) screening methods.
• To assess artificial intelligence (AI)-based tools, portable fundus cameras, and trained non-ophthalmologists in a developing country context.

Main Methods:

• A systematic literature search was performed across major databases (ScienceDirect, PubMed, Cochrane Library).
• Study quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies 2 tool.
• A meta-analysis of 21 AI-based studies was conducted to determine pooled sensitivity and specificity for DR detection.

Main Results:

• The meta-analysis included 25 studies, with 21 focusing on AI tools.
• Pooled sensitivity and specificity for AI in detecting any DR were 0.890 and 0.900, respectively.
• AI demonstrated high accuracy for referable DR (0.933 sensitivity, 0.903 specificity) and vision-threatening DR (0.891 sensitivity, 0.936 specificity).
• Portable fundus cameras and general physicians showed good agreement with gold standards.

Conclusions:

• AI-assisted DR screening shows significant potential in resource-limited settings.
• Portable fundus cameras and task-shifting to trained non-specialists complement AI, enhancing DR screening accessibility.