A novel approach for detection of diabetes patients using machine learning techniques

Abstract

Supporting medical decisions with data mining techniques allows for more efficient detection and treatment of disease while reducing the burden on doctors. Using data mining methods to diagnose diabetes in advance. Diseases of the kidney, eye, and heart are all linked to diabetes mellitus, which has the fourth highest fatality rate of any disease in the world. As a result, many people today suffer from the effects of this illness. This is where the bulk of current study focuses. Using data mining techniques aids in making accurate medical diagnoses. Different parts of the body are impacted by the chronic disease diabetes mellitus. The ability to accurately predict the spread of illnesses early on has the potential to save lives and give us command of them. Historically, diabetes has been diagnosed using a battery of physical exams; however, these procedures have been shown to be inaccurate. This study predicts the condition by utilising a variety of Data Mining techniques for the purpose of both predicting and diagnosing diabetes mellitus, therefore overcoming this constraint. Naive Bayes, K-Nearest Neighbor, Support Vector Machine, and Decision Tree are the most popular data mining methods. Seven hundred and sixty-eight occurrences from the Pima Indian Diabetes Dataset were used to create this dataset. Diabetic risk is scored and assigned to one of three categories: mild, moderate, and severe. In addition, the effectiveness of various algorithms for diabetes diagnosis has been analysed using this data. The outcomes obtained demonstrate how effective our classification method.