Enhanced Diabetes Prediction using Random Forest and XG Boost Machine Learning Classifiers with Dual Datasets

Abstract

Diabetes is a widespread chronic health condition with significant global implications. Early and accurate prediction of diabetes can enable timely interventions and improve patient outcomes. This paper explores the use of Random Forest and XG Boost machine learning classifiers to predict diabetes based on two distinct datasets. The first dataset includes attributes such as Pregnancies, Glucose levels, Blood Pressure, Skin Thickness, Insulin levels, BMI (Body Mass Index), Diabetes Pedigree Function, and Age. The Random Forest classifier achieves an accuracy of 91%, while the XG Boost classifier demonstrates superior performance with an accuracy of 93% in predicting diabetes on this dataset. The second dataset consists of attributes related to Hypertension, Heart Disease, Smoking History, BMI, HbA1c_level (glycated hemoglobin level), Blood Glucose Level, Diabetes Pedigree Function, and Age. In this dataset, the Random Forest classifier attains an accuracy of 96.98%, and the XG Boost classifier outperforms with an impressive accuracy of 97.25% in predicting diabetes. These results highlight the effectiveness of Random Forest and XG Boost machine learning classifiers in diabetes prediction, with the latter showing particularly promising results in both datasets. Such predictive models can assist healthcare professionals in identifying individuals at risk of diabetes, thereby enabling early intervention and better disease management.