Harnessing Machine Learning and Artificial Intelligence to Predict Maternal and Neonatal Health Outcomes
Code: G-1674
Authors: Fatemeh Zarei ℗, Zohre Ghaem Maghami *
Schedule: Not Scheduled!
Tag: Intelligent Virtual Assistant
Download: Download Poster
Abstract:
Abstract
Background and aim :As global initiatives to enhance maternal and neonatal health gain momentum, the integration of machine learning (ML) and artificial intelligence (AI) into clinical practice emerges as a pivotal advancement. This review synthesizes recent findings on the application of ML and AI in predicting maternal health outcomes, such as postpartum hemorrhage (PPH), perinatal depression (PND), and intrahepatic cholestasis of pregnancy (ICP), while also addressing their role in predicting neonatal mortality. Background: Postpartum complications, including hemorrhage and mental health disorders, pose significant challenges to maternal and neonatal well-being. Accurate early prediction is essential for timely intervention. PPH remains a leading cause of maternal morbidity and mortality globally, especially in low-resource settings. This review evaluates the effectiveness of various ML algorithms in predicting critical maternal health outcomes and analyzes recent advancements in AI-driven mortality prediction models for both mothers and neonates. Methods: A systematic literature search across five major scientific databases yielded 671 publications from the past decade. After screening, 18 studies were selected for in-depth analysis, focusing on methodologies and features of ML models, particularly Random Forest, in predicting PPH and neonatal mortality. Results: - Postpartum Hemorrhage (PPH): ML models showed a 95% higher likelihood of predicting PPH compared to traditional methods. Random Forest algorithms achieved the highest accuracy with a mean absolute error of 21.7. Key predictors included maternal age, gestational week, and cesarean delivery history. - Perinatal Depression (PND): The predictive model identified at-risk women based on mood status, previous depressive episodes, and sleep quality, demonstrating strong performance. - Intrahepatic Cholestasis of Pregnancy (ICP): The CatBoost model achieved an AUC of 0.9614, indicating excellent predictive capability for ICP severity. - Neonatal Outcomes: Key predictors for neonatal mortality included birth weight, gestational age, Apgar score, and gender. AI models enhanced risk assessment and facilitated early interventions. Conclusion : Machine learning and artificial intelligence have significant potential in predicting maternal and neonatal health outcomes. Integrating these techniques into clinical practice can enhance detection and interventions. Further research is essential for optimizing resources in this field.
Keywords
Machine Learning, Artificial Intelligence, Maternal Health
