Optimizing Type 2 Diabetes Classification with Feature Selection and Class Balancing in Machine Learning - Research Article

Aug 5, 2026

Type 2 Diabetes (T2DM) drives patient outcomes and treatment effectiveness, yet machine-learning detection systems often suffer from medical-data class imbalance and high feature dimensionality that reduce both accuracy and efficiency. This study integrates feature selection with imbalance handling to strengthen T2DM classification. Multiple feature selection methods are evaluated alongside SMOTE-based strategies, and several classifiers are compared using confusion-matrix metrics (accuracy, precision, recall) and runtime. Results show that combining informative feature selection with imbalance strategies substantially improves predictive performance, with IG+AdaBoost delivering consistently strong results.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages13

Reading Time33 min

AccessFree to download

Abstract

Introduction

Type 2 Diabetes background

Challenges: feature selection and class imbalance

Motivation for SMOTE and combined strategies

(Subsequent sections)

Optimizing Type 2 Diabetes Classification with Feature Selection and Class Balancing in Machine Learning Agus Wantoro*1, Aviv Fitria Yuliana2, Dwi Yana Ayu Andini3, Ikna Awaliyani4, Wahyu Caesarendra5 1Informatics, Universitas Aisyah Pringsewu, Indonesia 2,3Software Engineering, Universitas Aisyah Pringsewu, Indonesia 4Information Technology Education, Universitas Aisyah Pringsewu, Indonesia 5Departement of Mechanical and Mechatronics Curtin, University Malaysia [Email:](Email:1aguswantoro@aisyahuniversity.ac.id)[1](Email:1aguswantoro@aisyahuniversity.ac.id)[aguswantoro@aisyahuniversity.ac.id](Email:1aguswantoro@aisyahuniversity.ac.id) Received: Jul 23, 2025; Revised: Aug 8, 2025; Accepted: Aug 16, 2025; Published: Aug 24, 2025 Abstract | Type 2 Diabetes (T2DM) is a crucial factor in patient survival and treatment effectiveness. Errors in diabetes detection lead to disease severity, high costs, prolonged healing time, and a decline in service quality. Additionally, a major challenge in developing Machine Learning (ML)-based detection decision support systems is the class imbalance in medical data as well as the high feature dimensionality that can affect the accuracy and efficiency of the model. This research proposes an approach based on feature selection (FS) and handling class imbalance to improve performance in type 2 diabetes. Several feature selection techniques such as Information Gain (IG), Gain Ratio (GR), Gini Decrease (GD), Chi-Square (CS), Relief-F, and FCBF can perform feature selection based on weighting ranking. Furthermore, to address the imbalanced class distribution, we utilize the Synthetic Minority Over-Sampling Technique (SMOTE) . ML classification models such as Support Vector Machine (SVM), Gradient Boosting (GB), Tree, Neural Network (NN), Random Forest (RF), and AdaBoost were tested and evaluated based on the confusion matrix including accuracy, precision, recall, and time. The experimental results show that the combination of strategies for handling imbalanced classes significantly improves the predictive performance of ML algorithms. In addition, we found that the combination of feature selection techniques IG+AdaBoost consistently demonstrates optimal performance. This study emphasizes the importance of data preprocessing and the selection of the right algorithms in the development of machine learning-based T2DM detection systems. Accurate detection can reduce the severity of disease, lower treatment costs, speed up the healing process, and improve healthcare services.
Keywords: Diabetes, Feature selection, Imbalance class, Machine Learning. | | --- | | This work is an open access article and licensed under a Creative Commons Attribution-Non Commercial
4.0 International License
| 1. INTRODUCTION Diabetes Mellitus (DM) is one of the chronic diseases whose prevalence continues to increase globally, including in Indonesia. DM is a chronic metabolic disorder characterized by increased blood sugar levels in the body. This is caused by a disruption in insulin secretion. In 2023, approximately 415 million people aged between 20 and 79 years are reported to suffer from DM [1] . Diabetes mellitus (DM) is generally categorized into three: Type 1, Type 2, and Gestational Diabetes (GDM) . Type 1 diabetes (T1DM) affects 5% to 10% . This is characterized by autoimmune damage to the insulinproducing beta cells in the pancreas. [2] . Type 2 diabetes (T2DM) accounts for about 90% of all diabetes cases. In T2DM, the response to insulin is reduced. T2DM is mainly seen in people over the age of 45. It is increasingly observed in children, adolescents, and adults due to rising body weight, lack of physical activity, and a high-energy diet [3] Early detection and accurate diagnosis of T2DM is very important to prevent serious complications that can arise from this disease [4] . In recent years, rapid advances in the field of Machine Learning (ML) have paved the way for the development

FAQs

Why is early detection of Type 2 Diabetes important in this study?

Early detection and accurate diagnosis help prevent serious complications and support better treatment outcomes. The study frames ML-based detection systems as a way to improve predictive capability.

What two key challenges does the paper address for ML-based T2DM classification?

The paper focuses on feature selection (to reduce irrelevant or noisy variables) and class imbalance (where the minority class is underrepresented).

How do feature selection and SMOTE improve classification performance?

Feature selection methods identify informative subsets to improve efficiency and reduce overfitting, while SMOTE increases representation of the minority class to reduce bias toward the majority class. Together they improve performance across evaluated models.

Aug 5, 2026

Upload by Theodora