Diabetic Type Classification using Supervised Machine Learning Approaches

Abstract

Diabetic Retinopathy (DR) is the leading cause of blindness worldwide and a serious diabetic complication. To prevent vision loss, DR lesion diagnosis and categorization must be done early. DR early detection and treatment can significantly reduce the risk of vision loss. This paper focuses on classifying a sample into diabetic and non-diabetic using a variety of techniques, including Decision Tree, ANN, KNN, SVM, Random Forest, and Gradient Boosting Algorithms. The NCSU Diabetes the data set is pre-processed, and examples are trained and evaluated for accuracy; SVM and ANN achieve over 80% accuracy, demonstrating their potential in diabetes type classification. The PIMA Indians Dataset is used as a reference. The DR's manual diagnosing procedure Ophthalmologists' retina fundus scans take a lot of time, effort, money, and are prone to in contrast to computer-aided diagnosis systems, to misdiagnosis. Machine learning has recently been one of the most widely used methods that has improved performance in several categories, for example. The best classifier for diabetic retinopathy is determined by SVM, Decision. This compares ANN classifiers, Tree, Logistic Regression, and k-Nearest Neighbors paper. Additionally, a study of the existing DR datasets has been conducted. Numerous difficult also covered are topics that need further research. The results of comparing various machine learning algorithms with earlier studies are favourable. This research enhances the diagnosis of diabetic retinopathy by demonstrating the effectiveness of several machine learning classifiers and assisting in the creation of precise and effective computer-aided diagnostic tools for management and early detection.