Citation

Ali, Md Farman and Hasan, Kamrul and Haque, Md Ikramul and Nabi, Md Serajun and Tazwar, Md Tasin and Hossain, Mohammad Imtiaz and Noor, Jahid Hassan (2026) Early Detection of Fetal Distress using CTG and Machine Learning to Improve Maternal and Child Health. Journal of Engineering Technology and Applied Physics, 8 (1). p. 56. ISSN 2682-8383

Text View of Early Detection of Fetal Distress using CTG and Machine Learning to Improve Maternal and Child Health.pdf - Published Version Restricted to Repository staff only Download (3MB)

Abstract

This study addresses the critical challenge of fetaldistress identification, with either maternal or neonatal death coming first in this case. Manual CTG interpretation, being subjective, leads to disadvantages, and a high degree of inter-observer variability may result in misdiagnosis or late clinical intervention with detrimental consequences. Thus, our research fills the gap wherein there has been very little comprehensive and rigorous comparative analysis of a large number of machine learning models. We systematically studied a wide and diverse range ofclassifiers in our comparison, including tree-based ensembles such as XGBoost, Random Forest, CatBoost, and LightGBM, but also Gradient Boosting, Decision Tree, Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Logistic Regression, AdaBoost, Naive Bayes, and Convolutional Neural Networks (CNN). These models were trained and tested on a very large and open-source CTG dataset to validate their predictive power. Our findings reveal that the XGBoost model demonstrated superior performance with an impressive accuracy of 99.04%, while CatBoost, LightGBM, and Random Forest, which had intense predictive powers well above traditional diagnostic means. The above-mentioned accuracy-driven models have found their aptitude in capturing highly complicated nonlinear patterns occurring in CTG data and therefore hold a promising prospect to be developed and applied toward a large-scale and automated diagnostic aid. The successful implementation of these novel techniques would have huge potential for improving the quality of prenatal care and clinical decision-making in resource-poor areas, where expert supervision may be widely scarce.

Item Type:	Article
Uncontrolled Keywords:	Cardiotocography, fetal health
Subjects:	R Medicine > R Medicine (General) > R858-859.7 Computer applications to medicine. Medical informatics
Divisions:	Faculty of Artificial Intelligence & Engineering (FAIE)
Depositing User:	Ms Rosnani Abd Wahab
Date Deposited:	09 Jul 2026 04:47
Last Modified:	09 Jul 2026 04:47
URII:	http://shdl.mmu.edu.my/id/eprint/16369

Downloads

Edit (login required)