Diabetes Detection Using Machine Learning Algorithms on the Diabetes Dataset

Abstract

Diabetes mellitus is a chronic health condition affecting millions of people worldwide. Early detection of diabetes is crucial for effective treatment and management. With advancements in data science and machine learning (ML), it is now possible to analyze health data and predict the onset of diabetes in individuals. This report explores the application of various ML algorithms for diabetes detection using the Diabetes Dataset. Type 2 diabetes is major now a days. It arises when the body either fails to produce sufficient insulin or does not respond effectively to it, resulting in elevated blood sugar levels. If left unchecked, this condition can lead to severe health complications, including cardiovascular and renal diseases. This research focuses on supporting healthcare professionals in the accurate identification and diagnosis of Type 2 diabetes through the use of machine learning. A set of supervised learning algorithms was applied to build predictive models using eight key features from the Pima Indian Diabetes dataset. The study presents a detailed account of the methodology, experimental setup, and comparison with existing literature. Four widely used classification algorithms logistic regression, support vector machines (SVM), decision trees, and random forests were evaluated. Among them, SVM demonstrated superior performance with an accuracy of 77.27%, precision of 75.61%, recall of 51.38%, and a ROC-AUC score of 82.47%. The results indicate that our model outperforms an artificial neural network (ANN) model from earlier studies using the same dataset. This work aims to enhance the diagnostic process for Type 2 diabetes through machine learning and promote continued innovation in this domain.