EARLY DETECTION OF DIABETES MELLITUS USING MACHINE LEARNING MODELS.

This project focuses on the early detection of Diabetes Mellitus using machine learning models, specifically LightGBM (LGBM), XGBoost, and a hybrid Stacking Ensemble approach. The Stacking Ensemble combines the strengths of LightGBM and XGBoost as base models (Level 0), with Logistic Regression as the meta-learner (Level 1), improving the final prediction accuracy. The dataset used includes key features such as Pregnancies, Glucose, Blood Pressure, Skin Thickness, Insulin, BMI, Diabetes Pedigree Function, Age, and the target variable Outcome. Our results demonstrate that the Stacking Ensemble achieves the highest accuracy of 97.2%, surpassing individual model performances (LightGBM at 96.5%, XGBoost at 96.8%). A Flask-based application is developed to provide modules including Home, Register, Login, Classification, and Logout, ensuring an end-to-end solution for diabetes detection. The integration of multiple machine learning algorithms with an ensemble approach showcases a robust method for accurate early diagnosis of diabetes, highlighting the potential of machine learning in healthcare applications.

Keywords: LightGBM, XGBoost, Stacking Ensemble, Diabetes Mellitus, Machine Learning, Logistic Regression, Prediction, Flask, Healthcare