What Is the Role of AI in Diabetic Retinopathy Screening?

By Rasit Dinc

Diabetic retinopathy (DR) is a major complication of diabetes and a leading cause of preventable blindness in working-age adults worldwide. With the global prevalence of diabetes projected to reach 700 million by 2045, the burden on healthcare systems for DR screening is immense [1]. Traditional screening methods, which rely on manual interpretation of fundus photographs by ophthalmologists, are time-consuming and often inaccessible in underserved areas. This has created a critical need for more efficient and scalable screening solutions. Artificial intelligence (AI) has emerged as a transformative technology with the potential to revolutionize DR screening, offering a path towards earlier detection and improved patient outcomes.

At the heart of this revolution are deep learning algorithms, particularly convolutional neural networks (CNNs). These algorithms are inspired by the human brain's ability to recognize patterns and can be trained on vast datasets of retinal images to identify the subtle signs of DR. By analyzing thousands of images, CNNs learn to detect characteristic lesions such as microaneurysms, hemorrhages, and exudates with a high degree of accuracy [2]. The initial layers of these networks detect simple features like edges and colors, while deeper layers identify more complex patterns, allowing for a nuanced and automated analysis of retinal images. This automated process not only speeds up the screening process but also has the potential to reduce the workload of ophthalmologists, allowing them to focus on patients who require treatment.

In the United States, the Food and Drug Administration (FDA) has cleared three autonomous AI systems for DR screening: IDx-DR, EyeArt, and AEYE Health. These systems have demonstrated impressive performance in clinical trials, with sensitivity and specificity rates exceeding 87% and 89%, respectively [1]. The EyeArt system, for instance, has shown a sensitivity of 96% and a specificity of 88% for detecting more than mild DR [3]. The growing adoption of these technologies is reflected in the increasing number of Medicare claims for AI-based DR screening, with over 15,000 claims filed since 2022 [1]. This indicates a growing trust in AI's ability to accurately and reliably screen for DR.

Despite the promising results, the widespread implementation of AI in DR screening faces several challenges. One of the primary concerns is the issue of liability. While the responsibility for a missed diagnosis is likely to fall on the device manufacturer, the legal framework is still evolving [1]. Another challenge is the potential for bias in AI algorithms. If an algorithm is trained on a dataset that is not representative of the broader population, it may not perform as well in certain demographic groups. Ensuring equitable performance across diverse populations is crucial for the ethical and effective deployment of these technologies. Furthermore, the cost-effectiveness and reimbursement models for AI screening are still being evaluated to ensure financial viability for clinics and healthcare systems.

In conclusion, AI holds immense promise for transforming diabetic retinopathy screening. By providing a highly accurate, efficient, and scalable solution, AI has the potential to improve access to care, enable earlier detection, and ultimately prevent vision loss for millions of people with diabetes. While challenges related to regulation, liability, and equity remain, ongoing research and development are paving the way for a future where AI is an indispensable tool in the fight against diabetic blindness. As these technologies continue to mature and integrate into clinical practice, we can expect to see a significant impact on patient care and a brighter future for individuals with diabetes.

References

[1] Rajesh, A. E., & Lee, A. Y. (2025). AI for DR screening: Where are we in 2025?. Retina Specialist. Retrieved from https://www.retina-specialist.com/article/ai-for-dr-screening-where-are-we-in-2025

[2] Lim, J. I., Regillo, C. D., Sadda, S. R., Ipp, E., Bhaskaranand, M., Ramachandra, C., & Solanki, K. (2022). Artificial Intelligence Detection of Diabetic Retinopathy: Subgroup Comparison of the EyeArt System with Ophthalmologists’ Dilated Examinations. Ophthalmology Science, 2(4), 100228. https://doi.org/10.1016/j.xops.2022.100228

[3] Kong, M., & Song, S. J. (2024). Artificial Intelligence Applications in Diabetic Retinopathy: What We Have Now and What to Expect in the Future. Endocrinology and Metabolism, 39(3), 416-424. https://doi.org/10.3803/EnM.2023.1913