Comparative analysis of different machine learning algorithms for diabetes identification

Abstract

In today's world, diabetes has become one of the deadliest diseases and at the same time the most common disease not only in India but across the globe. Today, people of all ages can develop diabetes, which is linked to lifestyle, genetics, stress, and age-related factors. Whatever the cause of diabetes and diabetes-related diseases, we can identify the disease with different machine-learning algorithms. In the proposed work, we used Support Vector Machine (SVM) and Naive Bayes machine learning algorithms, which are mathematical and statistical concepts that help identify potential opportunities to be affected by diabetes-related diseases. After the data is preprocessed, the features that affect the prediction are selected by performing forward and backward feature selection. By analyzing the dimensionality reduction method of Principal Component Analysis (PCA) after selecting specific features, Naive Bayes achieves a prediction accuracy of 75%, which is significant compared to the 73% accuracy of SVM. This article is about identifying type 1 and type 2 diabetes. Accurate diagnosis and classification of diabetes are essential for effective treatment and management. When the immune system of the body attacks and destroys the cells responsible for producing insulin in the pancreas, it results in an autoimmune disease known as Type 1 diabetes. Type 2 diabetes is a metabolic disorder that occurs when the body becomes insulin resistant or does not produce enough insulin. This review article discusses the various methods and techniques used to identify type 1 and type 2 diabetes, including clinical, biochemical, and genetic markers .