Diabetic Retinopathy Detection System Using Naïve Bayes Method

This research discusses the detection of Diabetic Retinopathy using the Naïve Bayes method. Diabetes mellitus is a chronic disease that can cause complications such as Diabetic Retinopathy. The Naïve Bayes method automatically identifies early signs of DR. The background of this study is the high rate of diabetes and the lack of knowledge about diabetes and the high cost of diabetic retinopathy treatment so that early detection is needed. Eye image data from diabetes patients is collected and processed with a Naïve Bayes model before training. This research aims to improve early detection of DR and the HOG method, as feature extraction is also used in image pre-processing. Data is divided into training and test data to train and evaluate the model. This study used a diabetic retinopathy dataset with 2929 training and 733 testing data. The Naive Bayes model achieved an accuracy of around 69.5% on training data and 62.1% on test data. Although these results reflect relatively good performance, there are cases where the detection results do not always correspond to the actual classes. Further evaluation and improvement of the model can improve detection accuracy. This study provides initial insight into the potential and limitations of the model for detecting diabetic retinopathy.