Efficacy Evaluation of Machine Learning (ML) Tools in Enhancing Predictive Accuracy for High-Risk Obstetric Outcomes: A Systematic Review and Meta-Analysis

Accurate early diagnosis of high-risk obstetric outcomes, such as Preeclampsia (PE), Gestational Diabetes Mellitus (GDM), and preterm birth, is critical for reducing maternal and neonatal morbidity and mortality. The objective of this study was to conduct a comprehensive, quantitative assessment of the effectiveness of Machine Learning (ML) models compared to conventional clinical tools in screening for these disorders. Methods: This study utilized a PRISMA-guided systematic review and meta-analysis of literature published up to the end of ۲۰۲۴. Searches across PubMed, Embase, and Web of Science focused on comparative studies where ML models (including SVM, CNN, and RNN) were applied for risk prediction. Data points such as AUC, Sensitivity, Specificity, and Odds Ratios were extracted. Statistical meta-analysis was performed where data homogeneity permitted. Results: Eighteen studies involving a total of ۶۸,۵۰۰ pregnancies met the inclusion criteria. For PE prediction, Deep Learning (DL) models yielded a mean AUC of ۰.۹۱, significantly exceeding standard clinical tools (AUC=۰.۷۵) (p<۰.۰۰۱). Furthermore, ML models identified preterm birth an average of three weeks earlier than conventional instruments. Discussion & Conclusion: The evidence strongly supports ML as a novel paradigm in obstetrics, significantly boosting predictive accuracy. Practical application promises more timely and personalized interventions. However, challenges related to external validation and the need for algorithmic transparency (Explainable AI) persist for robust clinical adoption in settings like Spain.

کلیدواژه ها:

Machine Learning ، Systematic Review ، High-Risk Pregnancy ، Predictive Modeling ، Midwifery ، Preeclampsia