AI systems demonstrate promising potential for DR screening, study finds

AI-assisted screening for diabetic retinopathy showed comparable sensitivity and specificity compared to manual screening. Image credit: AdobeStock/ipopba

A recent study has shown the potential role artificial intelligence (AI) systems could play in diabetic retinopathy (DR) screenings. Study authors, led by Hasan Nawaz Tahir, MS, of the Department of Community Medicine at both the College of Medicine, Dwadimi, Shaqra University in Shaqra, Saudi Arabia; and the Khyber Medical College Peshawar in Peshawar, Pakistan, found in their systematic review and meta-analysis that the systems were able to achieve high sensitivity and specificity values, particularly in un-dilated eye screenings.¹

“These results highlight that AI systems, especially in un-dilated eye conditions, show promise for clinical use with reliable sensitivity and specificity, but variation exist depending on the system and clinical setting,” the study authors stated.

Researchers conducted a literature search for AI and manual screening methods of DR using PubMed and Google Scholar in order to identify relevant studies to include in the research that was published between January 2015 and September 2024. A second search was then done on February 2025, in which 13 additional studies were added to make a total of 25 included studies. Those that were included if they were observational or validation, in additional to evaluating AI algorithms or manual screenings for DR in patients from 15 to 90 years old that were diagnosed with DR and reported sensitivity and specificity outcomes for either dilated or un-dilated eye conditions, according to researchers. Risk of bias was assessed using the QUADAS-2 tool for validation studies, the AXIS tool for cross-sectional studies, and the Newcastle-Ottawa Scale for cohort studies. They were also assessed for quality by 2 independent reviewers to evaluation selection bias, outcome/exposure assessment bias, follow up bias, measurement bias, sample representativeness, reporting bias, index test bias, reference standard bias, flow and timing bias, and ethical considerations bias. Studies that were excluded were those that did not report the outcomes of interest, if the author of the studies did not respond to interest queries, or if the full text was not available to the study authors.¹

In total, 613,690 images used for screening were included in the study. For un-dilated eyes, AI screening had a pooled sensitivity of 0.90 [95% CI: 0.85–0.94] and pooled specificity of 0.94 [95% CI: 0.91–0.96]. Manual screening images had a pooled sensitivity of 0.79 [95% CI: 0.60–0.91] and pooled specificity of 0.99 [95% CI: 0.98–0.99]. As for dilated eyes, the pooled sensitivity of AI screening is 0.95 [95% CI: 0.91–0.97] and pooled specificity is 0.87 [95% CI: 0.79–0.92], with dilated eyes with manual screening sensitivity is 0.90 [95% CI: 0.87–0.92] and specificity is 0.99 [95% CI: 0.99–1.00].¹

“AI-assisted screening for diabetic retinopathy shows comparable sensitivity and specificity compared to manual screening,” the study authors stated. “These results suggest that AI can be a reliable alternative in clinical settings, with increased early detection rates and reducing the burden on ophthalmologists.”

The study authors also made not of the heterogeneity in the studies included in terms of study settings, photographic protocols, and reference standards, with reference standards for manual grading differing across studies, among other limitations. The researchers also concluded that since some of the studies had a moderate risk of bias, which could lead to over- or under-estimation of accuracy.¹ “To ensure that AI systems are safe and effective for real-world use, evaluations need to be conducted in representative clinical settings. Systems should be tested on a wide range of image qualities, and medical settings,” the study authors noted.

Reference:

1. Tahir HN, Ullah N, Tahir M, et al. Artificial intelligence versus manual screening for the detection of diabetic retinopathy: a comparative systematic review and meta-analysis. Front Med. 2025;12. https://doi.org/10.3389/fmed.2025.1519768