Citation

Tusher, Ekramul Haque and Akib, Md Abdullah and Rabbi, Uiadul Islam and Chowdhury, Mohammad Ziauddin and Othman, Khair Razlan and Khan, Ferose (2025) Ensemble Machine Learning Methods for Diabetes Prediction using Class Imbalance Technique. In: 2025 Multimedia University Engineering Conference, MECON 2025, 21 July 2025 - 23 July 2025, Cyberjaya, Malaysia.

Text 80.pdf - Published Version Restricted to Repository staff only Download (1MB)

Abstract

Millions of people throughout the world are living with diabetes mellitus, a metabolic illness that causes persistently high blood glucose levels. In order to intervene and control diabetes effectively, early and precise diabetes prediction is essential. Therefore, this study investigates the efficacy of various Machine Learning (ML) approaches for predicting diabetic illness using a comprehensive dataset containing medical and demographic information. In this study, an ensemble of state-of-the-art algorithms, including Random Forest (RF), Decision Trees (DT), XGBoost (XGB), Gradient Boosting (GB), AdaBoost (AB), and Light Gradient-Boosting Machine (LGB), was employed. Model evaluation using measures like F1-score, accuracy, precision, and recall was part of the methodology, which also included data pretreatment and feature selection. To combat class imbalance and guarantee fair model training, the Synthetic Minority Over-sampling Technique (SMOTE) was used. Results demonstrated the superior performance of ensemble and boosting techniques over individual decision tree-based models. With a 97.26% accuracy, a 98.90% precision, and an 84.69% F1-score, the LGB model was the best performer. With an F1-score of 81.09% and an accuracy of 97.24%, AB was right behind. Evaluation metrics, including ROC curves and AUC values, further validated the effectiveness of these models. By determining which ML models are most efficient in diabetes prediction, this research adds to the expanding corpus of knowledge in the field. The findings can potentially aid healthcare professionals in early disease forecasting.

Item Type:	Conference or Workshop Item (Paper)
Uncontrolled Keywords:	Diabetes prediction, ensemble learnin
Subjects:	R Medicine > R Medicine (General) > R855-855.5 Medical technology
Divisions:	Faculty of Engineering and Technology (FET)
Depositing User:	Ms Rosnani Abd Wahab
Date Deposited:	18 Mar 2026 07:58
Last Modified:	19 Mar 2026 00:33
URII:	http://shdl.mmu.edu.my/id/eprint/15566

Downloads

Edit (login required)