Millions of individuals worldwide are impacted by diabetic retinopathy, which is a leading cause of blindness and vision loss. The early detection of DR is extremely important and machine learning approaches have shown promising results in this regard. The present study thus uses a DR dataset acquired from the publicly available UCI machine learning repository. The dataset is normalized using the StandardScalar technique and then feature extraction is performed using Principal Component Analysis to extract the most significant features. Further, firefly algorithm is implemented for dimensionality reduction. This dataset is fed into a DNN framework which generates superior results in comparison to the contemporary ML models. To make the findings of the model usable for clinical decision-making explainable AI approach is used to ensure transparency and interpretability.