Prediction of postpartum hemorrhage (PPH) using machine learning algorithms in a Kenyan population
Santosh Yogendra Shah1, Sumant Saxena1, Satya Pavitra Rani1
1CognitiveCare Inc., Milpitas, CA, United States.
Frontiers in Global Women'S Health
|August 14, 2023
Summary
Machine learning models can predict postpartum hemorrhage (PPH) risk using patient data. The Naïve Bayes model showed the best performance, identifying key risk factors for PPH in Kenya.
Area of Science:
- Obstetrics and Gynecology
- Medical Informatics
- Public Health
Background:
- Postpartum hemorrhage (PPH) is a leading cause of maternal mortality globally, especially in low- and middle-income countries.
- Effective prediction models are crucial for early identification of at-risk women and timely intervention to reduce maternal morbidity and mortality.
Purpose of the Study:
- To develop and evaluate machine learning models for predicting postpartum hemorrhage (PPH).
- To utilize antenatal, intrapartum, and postnatal visit data from the Kenya Antenatal and Postnatal Care Research Collective cohort for PPH prediction.
Main Methods:
- Four machine learning models (logistic regression, naïve Bayes, decision tree, random forest) were trained and validated.
- Models were fine-tuned using feature selection (extra tree classifier), and performance was assessed using accuracy, sensitivity, and AUC-ROC.
Main Results:
- The Naïve Bayes model achieved the highest performance with 0.95 accuracy, 0.97 specificity, and 0.76 AUC.
- Seven factors including anemia, limited prenatal care, hemoglobin levels, pallor, and intrapartum blood pressure/respiratory rate were associated with PPH prediction.
Conclusions:
- Machine learning models show significant potential for predicting PPH in the Kenyan population.
- Further research with larger datasets is recommended to enhance prediction accuracy and inform personalized obstetric care, resource allocation, and maternal mortality reduction.
