Related Experiment Video
Updated: May 24, 2025
08:01
In utero Measurement of Heart Rate in Mouse by Noninvasive M-mode Echocardiography
Published on: November 22, 2013
11.4K
Proposing a machine learning-based model for predicting nonreassuring fetal heart
Nasibeh Roozbeh1, Farideh Montazeri1, Mohammadsadegh Vahidi Farashah1
1Mother and Child Welfare Research Center, Hormozgan University of Medical Sciences, Bandar Abbas, Iran.
Scientific Reports
|March 6, 2025
Summary
Machine learning models can predict nonreassuring fetal heart (NFH) conditions. Random forest classification demonstrated the highest performance, offering potential for improved perinatal care.
Area of Science:
- Obstetrics and Gynecology
- Medical Informatics
- Machine Learning in Healthcare
Background:
- Predicting nonreassuring fetal heart (NFH) patterns is critical for reducing perinatal complications.
- Limited research exists on identifying key predictors for NFH.
Purpose of the Study:
- To evaluate the efficacy of machine learning (ML) models in predicting NFH.
- To identify demographic, obstetric, maternal, and neonatal factors associated with NFH.
Main Methods:
- Retrospective analysis of singleton births (≥28 weeks gestation) from the Iranian Maternal and Neonatal Network (January 2020 - January 2022).
- Four ML models (Decision Tree, Random Forest, Extreme Gradient Boost, k-NN) were developed and compared.
- Chi-Square test identified potential NFH predictors (p < 0.05).
- Model performance was assessed using AUROC, accuracy, precision, recall, and F1 score.
Main Results:
- The incidence of NFH was 9.2%.
- NFH was associated with intrauterine growth restriction, late/post-term/preterm births, preeclampsia, placental abruption, primiparity, induced labor, and male fetus; doula support was associated with lower incidence.
- Random Forest (AUROC: 0.77) and k-NN (AUROC: 0.77) showed the best performance, with Random Forest achieving 0.77 accuracy and 0.72 precision.
Conclusions:
- Machine learning models, particularly Random Forest, show promising performance in predicting NFH.
- Further research is warranted to solidify the role of ML in predicting NFH and improving perinatal outcomes.
