Machine Learning based Early Predication and Detection of Diabetes Mellitus

According to a recent survey, diabetes mellitus has become the driving reason for death worldwide. Diabetes mellitus is critical disorders or illness that causes the imbalance of blood glucose levels due to the malfunction of the pancreas, which cannot make insulin hormones in the body; it is a silent and slow killer. It has a high risk for damage, malfunction and failure of human organs like the Kidney, heart, eye and nerve systems. There are several researches for the prediction and detection of Diabetes mellitus. The remarkable public health care infrastructure development for collecting critical and sensitive data. Many intriguing applications of machine learning algorithms are known for their ability to predict and identify diseases early on. The objective of this study is used to develop an effective early detection and prediction of diabetes mellitus using machine learning (ML) algorithms. The ML algorithms were performed on the Pima India Diabetes Dataset (PIDD) to develop the model. The proposed algorithm helps the practitioner to determine the seriousness of diabetes that is useful for the practitioner for the early recommendation of medicine, workout and treatment for the curing of disease. We developed many machine learning models in the experiment, such as Naïve Byase, KNN, Logistic Regression, SVM, decision tree, Random Forest, LightGBM and XGBoost to detect Type-II diabetes mellitus. The improved accuracy table shows the results of various models, and out of that, the XGBoost algorithm gives the highest accuracy of 89.07%.