孕前保健被证实可有效降低不良妊娠结局发生率，但有效措施十分匮乏。解决这一问题的关键是建立合适的孕前健康风险评估模型，对孕前健康状态及妊娠风险进行辨识评估，指导有效方案的制定。.本研究拟1）设计前瞻性巢式病例对照试验，对前期已有完整孕前健康信息记录的育龄妇女进行随访，采集其妊娠结局信息，运用逻辑回归法探索孕前健康状态的中医辨识结果与不良妊娠结局之间的相关性；2）结合本课题数据特点及研究目标，对logisitic回归与RBF神经网络模型进行比较分析，赋予二者合适的权重值，构建基于Logistic回归-RBF神经网络组合的孕前健康风险辨识评估模型；3）对该模型的预测准确率、捕获性和结果可理解性进行评价。.本研究的目标是在中医优生思想指导下，建立孕前健康风险辨识评估模型，通过对育龄妇女孕前健康状态的中医辨识，评估其发生不良妊娠结局的风险，为探索适合我国国情的孕前保健模式提供新工具。

Preconception care is considered as effective to reduce the rate of adverse pregnancy outcome, while effective method is still rare. To solve this problem, the key is establishing proper risk model, to evaluate the health status and risk before pregnancy.. This study will 1) follow up the pregnancy outcome from the female cohort whose detail information befroe pregnancy being collected, using prospective nested case-control method; then explore the relationship between preconception health status and pregnancy outcome; 2) establish two risk models built by logistic regression and RBF-ANN separately, then compared and combined them properly; 3) evaluate the combined model by ROC curve, lift hart, and misdiagnosis rate. . This study aims to establish a new preconception risk model under TCM theory, by diagnose the health status before pregnancy, to evaluate the risk between and after pregnancy.