内科

目的基于中孕期临床数据分析孕妇发生自发性早产（SPB）的影响因素，并建立预测模型。方法回顾性分析1 051名孕妇的临床资料，其中分娩时孕周<37周的孕妇为SPB组，≥37周的孕妇为足月组。使用多因素logistic回归模型探究孕妇发生SPB的影响因素。按照7∶3的比例将孕妇随机分成训练集和验证集，采用决策树算法建立孕妇发生SPB的预测模型，并采用受试者操作特征（ROC）曲线评估模型预测性能。结果多因素分析结果显示，分娩时年龄（OR=1.070，95%CI：1.001～1.144）、孕次（OR=1.888，95%CI：1.023～3.485），以及孕中期白细胞计数（OR=1.144，95%CI：1.026～1.276）、中性粒细胞与淋巴细胞比值（NLR）（OR=1.603，95%CI：1.152～2.232）、胎儿纤维连接蛋白（fFN）（OR=6.961，95%CI：3.740～12.955）、阴道清洁度（OR=6.673，95%CI：3.661～12.161）均是孕妇发生SPB的影响因素（均P<0.05）。训练集与验证集的决策树模型ROC曲线下面积分别为0.796（95%CI：0.720～0.871）和0.786（95%CI：0.658～0.913），准确度分别为93.99%和94.83%。Delong检验显示，验证集决策树模型ROC曲线下积与训练集决策树模型差异无统计学意义（D=0.126，P=0.786），提示模型预测效能较好。结论分娩时年龄、孕次，以及孕中期白细胞计数和NLR水平、fFN、阴道清洁度均是孕妇发生SPB的影响因素，基于这些因素构建的决策树模型，预测性能较好，可为临床实现孕妇发生SPB风险的个性化的预测提供参考。

Objective To analyze influencing factors for spontaneous preterm birth (SPB) in pregnant women based on the clinical data in the second trimester of pregnancy, according to which to build a prediction model. Methods The clinical data of 1,051 pregnant women were retrospectively analyzed, among them, pregnant women who delivered at gestational week <37 were the SPB group, and pregnant women who delivered at gestational week ≥ 37 were the full-term group. The multivariate logistic regression model was used to explore influencing factors for SPB in pregnant women. Pregnant women were randomly divided into the training set and the validation set in a ratio of 7∶3. The decision tree algorithm was used to establish a prediction model for SPB in pregnant women, whose performance was evaluated by the receiver operating characteristic (ROC) curve. Results The results of multivariate analysis showed that the age at the delivery (OR=1.070, 95%CI: 1.001~1.144) and times of pregnancy (OR=1.888, 95%CI: 1.023~3.485), as well as white blood cell count (OR=1.144, 95%CI: 1.026~1.276), neutrophil-to-lymphocyte ratio (NLR) (OR=1.603, 95%CI: 1.152~2.232), fetal fibronectin (fFN) (OR=6.961, 95%CI: 3.740~12.955), and vaginal clearing degree (OR=6.673, 95%CI: 3.661~12.161) in the second trimester of pregnancy, were influencing factors for SPB in pregnant women (all P<0.05). The areas under the ROC curves of the decision tree model in the training set and validation set were 0.796 (95%CI: 0.720~0.871) and 0.786 (95%CI: 0.658~0.913), respectively, and the accuracy rates were 93.99% and 94.83%, respectively. The Delong test results showed that there was no statistically significant difference in the area under the ROC curve between the decision tree models in the training set and the validation set (D=0.126, P=0.786), which indicated that the model had a good prediction performance. Conclusion Age at the delivery and times of pregnancy, as well as white blood cell count, NLR level, fFN, and vaginal clearing degree in the second trimester of pregnancy, are influencing factors for the occurrence of SPB in pregnant women, and the decision tree model based on these factors has a good prediction performance, which can provide a reference for the personalized prediction for the risk of SPB in pregnant women.