Pi Xinyu, Wang Junzhi, Chu Liangliang, Zhang Guochun, Zhang Wenli
School of Nursing, Shandong First Medical University (Shandong Academy of Medical Sciences), Jinan, 250117, Shandong, China.
Department of Obstetrics, The First Affiliated Hospital of Shandong First Medical University (Shandong Provincial Qianfoshan Hospital), Jinan, 250014, Shandong, China.
Sci Rep. 2025 May 4;15(1):15561. doi: 10.1038/s41598-025-00450-3.
This study explores the application of machine learning algorithms in predicting high-risk pregnancy among expectant mothers, aiming to construct an efficient predictive model to improve maternal health management. The study is based on the maternal health risk dataset (MHRD) from Bangladesh, covering multiple hospitals, community clinics, and maternal healthcare centers, and encompassing health data from 1014 pregnant women. Six machine learning algorithms-multilayer perceptron (MLP), logistic regression (LR), decision tree (DT), random forest (RF), eXtreme Gradient Boosting (XGBoost), and support vector machine (SVM)-are employed to construct predictive models. It is worth noting that MLP demonstrates superior performance compared with the other five algorithms. By applying the MLP method, the study successfully established an efficient pregnancy risk prediction model. The model evaluation results indicate that it has high accuracy in predicting pregnancy risks, with an overall accuracy rate of 82%, and particularly high accuracy in high-risk predictions, reaching 91%. With the computational support of an NVIDIA GPU RTX3050Ti, the model demonstrated excellent data processing capabilities, capable of predicting and processing 500 sets of data items per second. This study not only showcases the enormous potential of machine learning technology in the healthcare field, especially in the rapid and accurate identification of high-risk pregnancies, providing a powerful decision-support tool for medical professionals, but also offers significant reference value for future research in this area.
本研究探讨机器学习算法在预测孕妇高危妊娠中的应用，旨在构建一个高效的预测模型以改善孕产妇健康管理。该研究基于孟加拉国的孕产妇健康风险数据集（MHRD），涵盖多家医院、社区诊所和孕产妇保健中心，包含1014名孕妇的健康数据。采用六种机器学习算法——多层感知器（MLP）、逻辑回归（LR）、决策树（DT）、随机森林（RF）、极端梯度提升（XGBoost）和支持向量机（SVM）——来构建预测模型。值得注意的是，与其他五种算法相比，MLP表现出卓越的性能。通过应用MLP方法，该研究成功建立了一个高效的妊娠风险预测模型。模型评估结果表明，它在预测妊娠风险方面具有较高的准确性，总体准确率为82%，在高危预测方面准确率尤其高，达到91%。在NVIDIA GPU RTX3050Ti的计算支持下，该模型展示了出色的数据处理能力，能够每秒预测和处理500组数据项。本研究不仅展示了机器学习技术在医疗保健领域的巨大潜力，特别是在快速准确识别高危妊娠方面，为医学专业人员提供了一个强大的决策支持工具，而且为该领域未来的研究提供了重要的参考价值。
