<p>This research introduces a Deep Learning Approach for early Detection of Diabetic Retinopathy (DEDDR), leveraging the Pyspark computational framework. DEDDR addresses challenges in manual evaluation, achieving an exceptional accuracy rate of 99.9% in grading diabetic retinopathy. It incorporates advanced techniques, including image preprocessing and nerve segmentation. Strategic class weight balancing ensures equitable model training in the presence of class imbalance. The Retinoactive activation function, a unique creation, facilitates learning from negative values. Rigorous assessment metrics validate DEDDR’s robust capabilities. Notably, a customized loss function further refines model performance. The proposed approach carries profound implications for healthcare, accelerating early diabetic retinopathy detection and averting irreversible vision loss. Ongoing and future work encompasses model optimization, interpretability techniques, and seamless integration into clinical practice for real-world validation and impact assessment.</p>