Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhancing Metabolic Syndrome Prediction with Hybrid Data Balancing and Counterfactuals

Sanyam Paresh Shah, Abdullah Mamun, Shovito Barua Soumma, Hassan Ghasemzadeh

TL;DR

This work tackles the challenge of predicting MetS in the presence of class imbalance and data scarcity by systematically evaluating multiple ML models under standard preprocessing and a spectrum of data-balancing techniques. It introduces MetaBoost, a novel hybrid framework that combines SMOTE, ADASYN, and CTGAN with optimized weighting to improve predictive performance, achieving an accuracy of approximately 0.871 and an F1 of 0.868. A probabilistic analysis identifies elevated glucose and triglycerides as the strongest predictors, while counterfactual explanations quantify actionable feature changes needed to shift risk, offering clinical interpretability. The study demonstrates the value of hybrid data balancing and counterfactual analysis for robust, interpretable MetS risk prediction and provides a scalable approach for imbalanced health data more broadly.

Abstract

Metabolic Syndrome (MetS) is a cluster of interrelated risk factors that significantly increases the risk of cardiovascular diseases and type 2 diabetes. Despite its global prevalence, accurate prediction of MetS remains challenging due to issues such as class imbalance, data scarcity, and methodological inconsistencies in existing studies. In this paper, we address these challenges by systematically evaluating and optimizing machine learning (ML) models for MetS prediction, leveraging advanced data balancing techniques and counterfactual analysis. Multiple ML models, including XGBoost, Random Forest, TabNet, etc., were trained and compared under various data balancing techniques such as random oversampling (ROS), SMOTE, ADASYN, and CTGAN. Additionally, we introduce MetaBoost, a novel hybrid framework that integrates SMOTE, ADASYN, and CTGAN, optimizing synthetic data generation through weighted averaging and iterative weight tuning to enhance the model's performance (achieving up to a 1.87% accuracy improvement over individual balancing techniques). A comprehensive counterfactual analysis is conducted to quantify the feature-level changes required to shift individuals from high-risk to low-risk categories. The results indicate that blood glucose (50.3%) and triglycerides (46.7%) were the most frequently modified features, highlighting their clinical significance in MetS risk reduction. Additionally, probabilistic analysis shows elevated blood glucose (85.5% likelihood) and triglycerides (74.9% posterior probability) as the strongest predictors. This study not only advances the methodological rigor of MetS prediction but also provides actionable insights for clinicians and researchers, highlighting the potential of ML in mitigating the public health burden of metabolic syndrome.

Enhancing Metabolic Syndrome Prediction with Hybrid Data Balancing and Counterfactuals

TL;DR

This work tackles the challenge of predicting MetS in the presence of class imbalance and data scarcity by systematically evaluating multiple ML models under standard preprocessing and a spectrum of data-balancing techniques. It introduces MetaBoost, a novel hybrid framework that combines SMOTE, ADASYN, and CTGAN with optimized weighting to improve predictive performance, achieving an accuracy of approximately 0.871 and an F1 of 0.868. A probabilistic analysis identifies elevated glucose and triglycerides as the strongest predictors, while counterfactual explanations quantify actionable feature changes needed to shift risk, offering clinical interpretability. The study demonstrates the value of hybrid data balancing and counterfactual analysis for robust, interpretable MetS risk prediction and provides a scalable approach for imbalanced health data more broadly.

Abstract

Metabolic Syndrome (MetS) is a cluster of interrelated risk factors that significantly increases the risk of cardiovascular diseases and type 2 diabetes. Despite its global prevalence, accurate prediction of MetS remains challenging due to issues such as class imbalance, data scarcity, and methodological inconsistencies in existing studies. In this paper, we address these challenges by systematically evaluating and optimizing machine learning (ML) models for MetS prediction, leveraging advanced data balancing techniques and counterfactual analysis. Multiple ML models, including XGBoost, Random Forest, TabNet, etc., were trained and compared under various data balancing techniques such as random oversampling (ROS), SMOTE, ADASYN, and CTGAN. Additionally, we introduce MetaBoost, a novel hybrid framework that integrates SMOTE, ADASYN, and CTGAN, optimizing synthetic data generation through weighted averaging and iterative weight tuning to enhance the model's performance (achieving up to a 1.87% accuracy improvement over individual balancing techniques). A comprehensive counterfactual analysis is conducted to quantify the feature-level changes required to shift individuals from high-risk to low-risk categories. The results indicate that blood glucose (50.3%) and triglycerides (46.7%) were the most frequently modified features, highlighting their clinical significance in MetS risk reduction. Additionally, probabilistic analysis shows elevated blood glucose (85.5% likelihood) and triglycerides (74.9% posterior probability) as the strongest predictors. This study not only advances the methodological rigor of MetS prediction but also provides actionable insights for clinicians and researchers, highlighting the potential of ML in mitigating the public health burden of metabolic syndrome.

Paper Structure

This paper contains 22 sections, 2 equations, 3 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Materials and Methods
  3. Dataset
  4. Data Preprocessing
  5. Model Selection and Evaluation
  6. Analysis of Oversampling Strategies
  7. Analysis of Class Imbalance Handling Methods
  8. Counterfactual Analysis
  9. Results
  10. Probability Analysis
  11. Prior Probability
  12. Likelihood Analysis
  13. Posterior Probabilities
  14. Accuracy Comparison
  15. Precision, Recall, and F1
  16. ...and 7 more sections

Figures (3)

  • Figure 1: The MetaBoost framework integrates advanced data augmentation techniques (CTGAN, ADASYN, SMOTE) to address class imbalance in metabolic syndrome prediction. The pipeline consists of oversampling, model training, risk prediction, and counterfactual analysis, enabling early intervention for at-risk individuals.
  • Figure 2: Comparison of accuracies of hybrid data balancing approaches across different weights. Weights: ($\lambda,1-\lambda$).
  • Figure 3: Decision boundary generated by random forest using 20 data points from the counterfactual examples. Here, Class 0 indicates an absence of MetS, and Class 1 indicates the presence of MetS.