Prediction of Type-2 Diabetes Mellitus Disease Using Machine Learning Classifiers and Techniques

Abstract

The technological advancements in today's healthcare sector have given rise to many innovations for disease prediction. Diabetes mellitus is one of the diseases that has been growing rapidly among people of different age groups; there are various reasons and causes involved. All these reasons are considered as different attributes for this study. To predict type-2 diabetes mellitus disease, various machine learning algorithms can be used. The objective of using the algorithm is to construct a predictive model to critically predict whether a person is affected by diabetes. The classifiers taken are logistic regression, XGBoost, gradient boosting, decision trees, ExtraTrees, random forest, and light gradient boosting machine (LGBM). The dataset used is PIMA Indian Dataset sourced from UC Irvine Repository. The performance of these algorithms is compared in reference to the accuracy obtained. The results obtained from these classifiers show that the LGBM classifier has the highest accuracy of 95.20% in comparison with the other algorithms.