• Home
  • Search
  • WHEN DATA AUGMENTATION HURTS: A SYSTEMATIC EVALUATION OF SMOTE-BASED TECHNIQUES ON MEDICAL DATASETS
  • Cite Icon1
  • https://doi.org/10.26483/ijarcs.v16i4.7313Copy DOI Icon

WHEN DATA AUGMENTATION HURTS: A SYSTEMATIC EVALUATION OF SMOTE-BASED TECHNIQUES ON MEDICAL DATASETS

  • Abstract
  • Literature Map
  • Citations
  • Similar Papers
Abstract

Data augmentation techniques, particularly Synthetic Minority Over-sampling Technique (SMOTE) and its variants, are routinely applied to address class imbalance in medical datasets. However, the assumption that augmentation universally improves classification performance remains largely unvalidated. This study presents a systematic evaluation of four SMOTE-based augmentation methods across three medical datasets to determine when these techniques help or harm model performance. The research evaluated SMOTE, ADASYN, BorderlineSMOTE, and SVM-SMOTE on breast cancer diagnosis, heart disease prediction, and diabetes detection datasets, representing varying levels of class imbalance (ratios: 1.17 to 2.02) and baseline performance (F1 scores: 0.667 to 0.966). Random Forest classifiers were employed with both standard and regularized configurations to ensure robust findings. Each augmentation method underwent rigorous evaluation through 10 independent runs with statistical significance testing and effect size analysis. Results revealed that augmentation significantly degraded performance on the high-performing Breast Cancer dataset, with all methods showing statistically significant decreases (p < 0.05) and F1 scores dropping by up to 2.2%. Conversely, the Pima Diabetes dataset, characterized by lower baseline performance and higher imbalance, showed improvements up to 4.76% with SVM-SMOTE. Heart Disease exhibited mixed results, with only ADASYN achieving meaningful improvement. Analysis uncovered a strong negative correlation (r = -0.997) between baseline model performance and augmentation effectiveness, providing a more reliable predictor than traditional class imbalance ratios. The study establishes an evidence-based decision framework: augmentation should be avoided when baseline F1 exceeds 0.95 or imbalance ratios fall below 1.5, considered for baseline F1 below 0.70 with imbalance ratios above 1.8, and carefully validated for intermediate cases. These findings challenge current practices of routine augmentation application and demonstrate that synthetic sample generation can blur decision boundaries in well-separated feature spaces. The research provides practitioners with validated guidelines for determining when augmentation techniques genuinely improve medical classifiers versus when they cause harm, ultimately supporting more effective development of clinical decision support systems.

Similar Papers
  • Research Article
  • Citations64

A multiple combined method for rebalancing medical data with class imbalances

  • May 31, 2021
  • Computers in Biology and Medicine
  • Yun-Chun Wang +1
  • Conference Article
  • Citations8

Modeling of class imbalance using an empirical approach with spambase dataset and random forest classification

  • Oct 13, 2014
  • Kiranmayi Kotipalli +1
  • Research Article
  • Citations6

Whale Optimization-based Synthetic Minority Oversampling Technique for Binary Imbalanced Datasets

  • Jan 01, 2024
  • Procedia Computer Science
  • Pooja Tyagi +2
  • PDF
  • Research Article
  • Citations24

Searching for Optimal Oversampling to Process Imbalanced Data: Generative Adversarial Networks and Synthetic Minority Over-Sampling Technique

  • Aug 21, 2023
  • Mathematics
  • Gayeong Eom +1
  • Research Article
  • Citations3

Improving remote sensing scene classification with data augmentation techniques to mitigate class imbalance

  • Oct 08, 2025
  • Frontiers in Computer Science
  • Ping Wang +3
  • Research Article
  • Citations30

Malicious web domain identification using online credibility and performance data by considering the class imbalance issue

  • Dec 04, 2018
  • Industrial Management & Data Systems
  • Zhongyi Hu +4
  • Research Article
  • Citations1

Fuzzy-driven broad learning system with class probability and density awareness for multi-view data.

  • Sep 01, 2026
  • Neural networks : the official journal of the International Neural Network Society
  • M Tanveer +4
  • Research Article
  • Citations1

Explainable machine learning for early heart disease risk prediction: Insights from a clinical dataset in Bangladesh

  • Mar 01, 2026
  • Intelligence-Based Medicine
  • Arpita Chakraborty +4
  • Research Article
  • Citations3

Enhanced Classification of Imbalanced Medical Datasets using Hybrid Data-Level, Cost-Sensitive and Ensemble Methods

  • Apr 22, 2024
  • International Research Journal of Multidisciplinary Technovation
  • Ayushi Gupta +1
  • Conference Article
  • Citations139

Intrusion Detection Using Random Forests Classifier with SMOTE and Feature Reduction

  • Nov 01, 2013
  • Abebe Tesfahun +1
  • Research Article

Evaluating The Effectiveness of Augmentation and Classifier Algorithms for Fraud Detection: Comparing CGAN and SMOTE with Random Forest and XGBoost

  • Oct 04, 2025
  • Applied Information System and Management (AISM)
  • Sarmini Sarmini +2
  • Research Article

Evaluating The Effectiveness of Augmentation and Classifier Algorithms for Fraud Detection: Comparing CGAN and SMOTE with Random Forest and XGBoost

  • Oct 07, 2025
  • Applied Information System and Management (AISM)
  • Sarmini Sarmini +2
  • Research Article
  • Citations136

SMOTE-LOF for noise identification in imbalanced data classification

  • Feb 09, 2021
  • Journal of King Saud University - Computer and Information Sciences
  • Asniar +2
  • Book Chapter

Classifier Learning from Imbalanced Corpus by Autoencoded Over-Sampling

  • Jan 01, 2019
  • Eunkyung Park +2
  • Research Article
  • Citations58

An Effective Cost-Sensitive XGBoost Method for Malicious URLs Detection in Imbalanced Dataset

  • Jan 01, 2021
  • IEEE Access
  • Shen He +4
Cactus Communications logo

Copyright 2026 Cactus Communications. All rights reserved.