Reducing severe adverse events for women during pregnancy through prediction modelling

产妇特异性早期预警评分（moews）是一种简单的工具，用于使用静态生理标记来预测女性病情恶化的可能性。然而，有证据表明，传统的风险评估可能低估了与怀孕有关的恶化的速度和深度，对这些措施的依赖导致了忽视迹象的情况。随着国民保健制度转向使用电子健康记录，这是一个理想的机会，可以开发强大的模型，及早预测孕产妇健康状况的恶化，并降低孕产妇发病率和死亡率。虽然存在产科预测模型，但很少有独立于其开发数据的评估（外部验证），在关键临床亚组中评估的更少。建议在开发新的预测模型之前，从外部验证并考虑更新现有的模型。尽管存在关于如何在开发过程中处理丢失数据的指导方针，但是在模型实现中如何处理单个丢失的数据存在相当大的差异。本研究的目的是确定现有moews在预测孕产妇死亡和孕产妇发病率方面的有效性和适用性，并研究如何改进现有方法。

Maternal-specific Early Warning Scores (MOEWSs) are simple tools designed to predict the likelihood of a women deteriorating using static physiological markers. However, evidence has shown the rapidity and depth of deterioration in relation to pregnancy can be under-predicted by traditional risk assessment, and reliance on these measures has led to instances where signs were ignored. As the NHS moves to the use of electronic health records, this is the ideal opportunity to develop robust models that can predict deteriorating maternal health early and reduce maternal morbidity and mortality.Although obstetric prediction models exist, very few have been evaluated on data independent of that on which they were developed (external validation), fewer have been assessed in key clinical subgroups. It is recommended to externally validate and consider updating an existing model before developing a new prediction model. Although guidelines exist on how to handle missing data during development, there is considerable variation in how to handle an individual's missing data at model implementation. The aim of this study is to determine the effectiveness and suitability of currently available MOEWSs at predicting maternal death and maternal morbidity and to investigate how current methods can be improved.