育龄期多囊卵巢综合征风险预测模型的构建

A nomogram model for predicting polycystic ovary syndrome

目的建立和验证基于多囊卵巢综合征(polycystic ovarian syndrome,PCOS)相关因素的列线图模型以预测多囊卵巢综合征的发生。方法收集2017-2019年某三甲医院生殖助孕中心的PCOS患者162例和妇产科健康备孕的女性162例,随机抽取70%分配到训练集,其余被分配到测试集。将参与者的基本信息、生活行为因素、实验室检测指标等采用LASSO回归进行筛选,建立多因素Logistic回归模型,利用列线图实现预测模型的可视化,并进行模型的验证。结果最终进入预测模型的是体质指数(body mass index,BMI)、高脂饮食、初潮年龄、月经周期、抗缪勒管激素(anti-Müllerian hormone,AMH)水平5个变量。训练集和测试集模型曲线下的面积分别为0.773(95%CI 0.712~0.833)和0.740(95%CI 0.642~0.839),分别用于预测和内部验证。结论基于PCOS相关因素建立预测PCOS风险的列线图模型具有良好的精准度和区分能力,有助于对PCOS患者进行自我风险评估和健康咨询。

Objective The early detection of polycystic ovary syndrome(PCOS)is very important for its prevention and treatment.Therefore,the purpose of this study was to establish and verify the nomogram model based on PCOS related factors to predict the occurrence of PCOS.MethodsThis study was an auxiliary analysis of cross-sectional studies.From 2017 to 2019,162 PCOS patients and 162 healthy pregnant women in obstetrics and gynecology in a“Three-A”hospital were collected,and 70%of them were randomly selected and assigned to the training,while the rest were assigned to the test group.The basic information,life behavior factors,laboratory test indicators and other information of participants were input into LASSO regression to screen the variables included in the model.The nomogram model was established by multivariable Logistic regression,and the model was verified.ResultsBody mass index(BMI),high fat diet,menarche age,menstrual cycle and anti-Müllerian hormone(AMH)level were finally entered into the prediction model.The areas under the curve of training group and validation group were 0.773(95%CI 0.712-0.833)and 0.740(95%CI 0.642-0.839),which were used for prediction and internal verification.ConclusionBased on the related factors of PCOS,a nomogram model for column prediction of PCOS was established in this study.The model has good accuracy and discrimination,which is helpful to self-risk assessment and health consultation of PCOS patients.