Study On Prediction Models Of Pregnancy Outcomes In PCOS Infertile Women Based On Clinical Randomized Controlled Trials

Object:The first purpose of this study is to evaluate the prediction model of PCOS pregnancy outcomes based on the stage established of prediction models,and observe its performance systematically.The second one is a clinical study of acupuncture and clomiphene combined treatment of anovulatory infertility women based on PCOS,using the common demographic data,laboratory examination and related history data to establish multivariate prediction model for prediction of ovulation,pregnancy,pregnancy,miscarriage and live birth outcome index,total population and at the same time respectively.Different people based on the prediction of pregnancy,stage specific predictors of pregnancy.Methods:1.The methods of evidence-based medicine were used to retrieve the literature on the prediction model of pregnancy outcomes of PCOS patients.The quality of the included literature,the stage of model establishment,the performance of the model and so on were evaluated,and the literature management was carried out using Endnote X8.2.The basic information,clinical data and pregnancy outcome information of 725 anovulation PCOS patients who met the diagnostic criteria and data integrity were collected.SAS 9.4 software was input into the database accurately and completely,and the data were statistically analyzed.Logistic stepwise regression method was used to select predictors related to pregnancy outcomes from candidate predictors.The significant level was defined as 0.1,and P value<0.1 was included in the final model.At the same time,the interaction of intervention measures and other included predictive factors is evaluated,and the P value<0.1 will also be included in the final model.The final prediction model is then determined according to the candidate prediction factors,age,intervention and interaction.The prediction model is shown as statistics,P values,OR values and corresponding 90%CI.After the model was established,the Hosmer-Lemeshow(H-L)test was used to evaluate the scale of the model area and to evaluate the calibration degree by using the ROC curve and the AUC.Result:1 Study on prediction models of pregnancy outcomes in PCOS patients1.1 A total of 16 articles were included in the analysis of the PCOS pregnancy outcomes prediction models,of which 14 reported a new prediction model,and 2 were tested internally,and the other 2 were external validation of the existing models.Most studies are prospective studies.The main diagnostic criteria of PCOS were the 2003 Rotterdam diagnostic criteria.The main intervention measures were CC or metformin.The prediction of pregnancy outcome is mainly ovulation,followed by pregnancy and live birth,without a predictive abortion model.1.2 Among the models reported in 14 papers,12 models were continuous and the data were collected prospectively.The description of the 11 models before diagnosis and treatment is sufficiently detailed;the description of the variables and pregnancy outcome in the 13 is sufficiently detailed.Only a few models reported missing data,of which only one filled the missing data.1.3 2 of the analysis papers were internally validated,and 2 were external validation of the existing models.In external validation,the performance of the original model is good.2 Prediction models of pregnancy outcomes in PCOS infertile women base on a clinical study2.1 In this study,a total of 9 multivariate prediction models,respectively,live births in the general population,pregnancy,ovulation,pregnancy abortion and live birth ovulation prediction model and population prediction model,the population of pregnant pregnant pregnancy and abortion and pregnancy prediction model,population forecast model.2.2 Predictive factors in the prediction model including clomiphene and acupuncture treatment,age,BMI score,acne,pulse pressure,MAP,E2,TT,Free-T,LH/FSH ratio,FAI,FPG,SHBG,LDL,AMH,the average menstrual cycle,previous pregnancy history and past history of abortion,pregnancy time,try smoking history,including BMI,E2 and age were also in clomiphene treatment interaction.The above predictors exist in at least one prediction model.2.3 Clomiphene in the general population prediction model,predict the success of ovulation,pregnancy,pregnancy and live birth,and the crowd in the ovulation prediction model and conception conception people pregnancy prediction model is risk factors;acupuncture treatment had no effect in the general population prediction model,and in pregnant women live births predictive model as a risk factor.AMH is a risk factor for prediction models of ovulation and pregnancy in the total population prediction model.LH/FSH ratio predicts low abortion rate in pregnant women and high live rate in the total population.Kaohsiung hormone related indicators,including TT/FAI,SHBG,and Free-T,have a negative impact on the various stages of pregnancy.2.4 9 multivariable prediction models AUC are 0.68-0.92,which indicates that the model has better calibration performance,and the H-L test is more than 0.05,suggesting that the model has better distinguishing performance.Conclusion:1 In the existing PCOS pregnancy outcome prediction models,two models were verified by external tests,indicating that the model has good prediction performance.These models can be used as a clinical guidance tool for guiding patients with similar models.2.1 Chinese anovulatory PCOS infertile women can use simple clinical data to establish a multivariate prediction model of pregnancy outcome,and it is also feasible to establish a model based on the establishment of a pregnancy model to explore the specific predictors of different stages of pregnancy.2.2 There are two interventions in this study.Clomiphene and acupuncture treatment affect the incidence of early abortion and late abortion,respectively.2.3 AMH mainly affects early pregnancy,while LH/FSH ratio mainly affects late pregnancy.The androgen hormone related indicators have negative effects on all stages of pregnancy.