Machine learning approaches towards risk assessment and prediction of adverse pregnancy outcomes

项目摘要 该项目的主要目标包括了解分子、遗传和生物学之间的相互作用。 与不良妊娠结局（APO）相关的临床因素， 在APO发生之前对其进行风险评估，并制定收集更多 高危受试者常规治疗的临床数据。为了实现这些目标，我们组建了一个团队， 具有临床、翻译和计算专业知识的研究人员能够识别新的 APO的贡献者，以及使用数据驱动和 理论上合理的机器学习方法。我们的战略将依靠先进的机器 学习以及临床，遗传和分子数据的整合，并有望带来 精准医学对怀孕期间和怀孕后妇女的治疗和经验。我们将 主要依赖于国家“未产妊娠结局研究： 监测准妈妈”;即，nuMoM 2b研究。使用10，038名未经产妇女的队列，我们 将有效地实现3个目标：整合遗传，临床和分子特征，以深入研究 了解APO;开发用于高级风险预测的机器学习模型;以及 为风险评估和模型开发积极收集数据。使用close 计算和临床科学家之间的合作，我们相信这一建议将导致 在理解APO的分子和临床方面以及评估 因此，我们必须采取措施，降低APO的风险，从而为孕产妇健康做出切实贡献。

PROJECT SUMMARY The primary objectives of this project include understanding the interplay between molecular, genetic and clinical factors related to adverse pregnancy outcomes (APOs), method development for accurate risk assessment of APOs well before they occur, and method development for collecting additional clinical data in routine treatment of at-risk-subjects. Towards these goals we have assembled a team of investigators with clinical, translational, and computational expertise capable of identifying novel contributors to APOs as well as facilitating clinician-patient interactions using data-driven and theoretically sound machine learning approaches. Our strategies will rely on advanced machine learning as well as integration of clinical, genetic, and molecular data and hold promise to bring precision medicine to the treatment and experience of women during and post pregnancy. We will predominantly rely on the data collected during the national “Nulliparous Pregnancy Outcomes Study: monitoring mothers-to-be”; i.e., the nuMoM2b study. Using the cohort of 10,038 nulliparous women, we will efficiently accomplish 3 Aims: to integrate genetic, clinical, and molecular features towards a deep understanding of APOs; to develop machine learning models for advanced risk prediction; and to engage in active data collection towards risk assessment and model development. Using a close collaboration between computational and clinical scientists, we believe this proposal will result in important advances in understanding the molecular and clinical aspects of APOs as well as assessing the risk for APOs and thus providing tangible contributions to maternal health.