Diabetic retinopathy (DR) is a significant complication of diabetes and a leading cause of preventable blindness globally, highlighting the crucial need for early detection to allow timely intervention and reduce vision loss. Despite advancements in imaging technologies, many existing detection methods still rely on manual interpretation, resulting in inconsistencies and delays. This study aims to address these challenges by developing an innovative hybrid system that leverages advanced machine learning techniques for enhanced analysis of Optical Coherence Tomography (OCT) image features through the retina using the APTOS Dataset. Our approach integrates traditional image processing with a modified Visual Geometry Group (VGG) architecture. It incorporates novel feature fusion mechanisms such as OCT feature, Deep Feature, and Clinical Feature leading to a remarkable accuracy of 94.3% while processing images in just 1.2 s. This performance signifies a substantial improvement over current methodologies, particularly in identifying early stage DR. By combining cutting-edge technology with clinical applications, this research contributes to the growing field of AI-driven medical diagnostics, aiming to improve patient outcomes and streamline DR detection.