Related Experiment Video
Updated: Sep 21, 2025
14:40
Instrumentation of Near-term Fetal Sheep for Multivariate Chronic Non-anesthetized Recordings
Published on: October 25, 2015
9.5K
HGSORF: Henry Gas Solubility Optimization-based Random Forest for C-Section prediction and XAI-based cause analysis
Md Saiful Islam1, Md Abdul Awal2, Jinnaton Nessa Laboni2
1Department of Computer Science, College of Computer and Information Sciences, King Saud University, Riyadh 11543, Saudi Arabia.
Computers in Biology and Medicine
|June 6, 2022
Summary
A new machine learning model, Henry gas solubility optimization-based random forest (HGSORF), accurately predicts cesarean or C-section (CS) delivery. This model improves upon existing methods, offering a 98.33% accuracy for CS prediction using the PDHS dataset.
Area of Science:
- Medical Informatics
- Machine Learning
- Public Health
Background:
- Unnecessary cesarean or C-section (CS) deliveries pose risks to maternal and neonatal health and incur significant costs.
- Existing machine learning models struggle to accurately predict CS delivery probability, necessitating improved predictive tools.
Purpose of the Study:
- To develop and evaluate a novel, stable predictive model for CS delivery using an improved Henry gas solubility optimization (HGSO)-based random forest (RF) algorithm, termed HGSORF.
- To enhance the accuracy of CS probability prediction compared to existing methods and classifiers.
Main Methods:
- Proposed the HGSORF model, integrating HGSO for RF hyperparameter tuning to avoid local minima and improve classification of CS and non-CS cases.
- Utilized the Pakistan Demographic and Health Survey (PDHS) dataset, applying the ADAptive SYNthetic (ADASYN) algorithm for data balancing.
- Compared HGSORF performance against Gaussian Naive Bayes (GNB), linear discriminant analysis (LDA), K-nearest neighbors (KNN), gradient boosting classifier (GBC), and logistic regression (LR).
- Employed eXplainable artificial intelligence (XAI) tools, including SHapely Additive exPlanations (SHAP) and Local Interpretable Model-Agnostic Explanations (LIME), for model interpretability.
Main Results:
- The HGSORF model achieved a superior accuracy of 98.33% on the PDHS dataset.
- HGSORF demonstrated better performance than other commonly used hyperparameter optimization algorithms for RF.
- XAI tools provided insights into the global and local factors influencing CS prediction.
Conclusions:
- The HGSORF model offers a highly accurate and interpretable solution for predicting cesarean delivery.
- The developed model can serve as a valuable decision support system for clinical staff, potentially reducing unnecessary CS rates.
- The study highlights the effectiveness of HGSO for optimizing RF hyperparameters in medical prediction tasks.
