AI-Based Early Detection of Diabetic Retinopathy to Prevent Severe Visual Impairment

Abstract

Diabetic retinopathy (DR) is a leading cause of preventable blindness among working-age adults globally, with a significant economic and social burden. Early detection plays a main role in preventing the progress of the disease to advanced stages that cause irreversible vision loss. Traditional screening methods, although effective, are often resource-intensive and time-consuming. This research introduces an AI-powered deep learning approach using Convolutional Neural Networks (CNN) to find and categorize diabetic retinopathy from retinal fundus images with high accuracy and minimal human intervention. The proposed system is trained and validated on a labeled dataset, incorporating advanced preprocessing technique and data augmentation, and the model achieved promising results in binary classification (DR vs. No DR), with robust evaluation through accuracy, recall, precision, F1-score, and ROC-AUC metrics. Results suggest that the developed CNN model can serve as an effective decision-support system for ophthalmologists, especially in under-resourced regions. Furthermore, the methodology can be extended to other retinal diseases and adapted for mobile diagnostic platforms.