Overview of the Scope of Application of Machine Learning Models in Predicting Maternal Delivery Risk and Outcomes

Childbirth is a normal physiological process for human reproduction, but there are various risks to maternal and fetal health and safety during this process . The 'Action Plan to Enhance Maternal and Infant Safety (2021-2025)' proposes a goal to reduce the national maternal mortality rate to 14.5 per 100,000 by 2025, requiring the implementation of pregnancy risk screening and assessment to improve the level of maternal and infant safety and lay a foundation for the 'Healthy China 2030' goal . In traditional childbirth models, women often experience numerous adverse experiences during labor, such as severe pain, fear, and anxiety, which can affect the progress of labor and maternal outcomes . Therefore, clinical practice must highly value interventions for adverse childbirth to predict risks and adverse outcomes during labor . Machine learning (machine learning, ML) is a significant branch of artificial intelligence, with its core lying in training algorithms through data to enable computers to perform predictive, classification, or decision-making tasks without explicit programming, allowing the construction of risk prediction models to estimate the probability of an individual developing a certain disease or the likelihood of future onset . ML models can integrate complex nonlinear relationships between multiple predictive variables, offering advantages over traditional statistical methods, especially in predicting early-onset diseases, where they demonstrate good accuracy and potential . In recent years, ML models and related research in the field of obstetrics have garnered significant attention. ML can improve the detection rate of risks during childbirth, reduce medical errors, and shorten diagnostic times . Therefore, this review aims to introduce ML algorithm types, data processing, application types, and application effects for predicting maternal childbirth risks and outcomes, deeply exploring the impact of ML model applications and effects on predicting maternal childbirth risks and outcomes, and providing references for the development and application of ML models for predicting maternal childbirth risks and outcomes in the future.