剖宫产术后阴道试产人群缩宫素智能调控模型的构建

目的：构建剖宫产术后阴道试产（VBAC）人群缩宫素（OT）智能调控模型，进一步探索产时OT用药的智能精准调控方案。方法：收集2014 年1月至2020 年5月于温州医科大学附属第一医院产科分娩，并在产程中使用OT的VBAC产妇相关资料，对电子病历资料运用多重线性回归法筛选建模变量，并结合胎心宫缩图提取的胎心、宫缩频率等变量，基于XGBoost算法建立OT调控模型，同时与Logistic回归模型和传统决策树进行比较，以准确率、查准率、召回率、F1值评价模型预测性能，并验证模型临床应用效果。结果：共纳入124例VBAC产妇，1 005条OT调节记录，XGBoost模型性能最佳，5折交叉验证下模型的准确率为0.82，查准率0.84，召回率0.80，F1值0.82，其中宫缩持续时间、宫腔压力、胎心、宫缩频率和前次剖宫产间隔时间是对建模贡献度较大的变量。临床验证表明预测模型精确度接近专家的决策水平，并优于低年资助产士。结论：本研究基于XGBoost构建OT调控模型，实现了产时OT的实时智能调控，其响应速度快、模型精度高、外推性强，对产科临床护理工作具有积极意义。

Construction of intelligent regulation model of oxytocin in vaginal birth after cesarean

Objective: To construct an intelligent regulation model of oxytocin (OT) in vaginal trial delivery of (VBAC) after cesarean section and to further explore the intelligent and precise regulation and control scheme for the use of OT during delivery. Methods: A cross-sectional study design was used to collect the data of VBAC parturients who delivered in the obstetrics department of the First Affiliated Hospital of Wenzhou Medical University from January 2014 to May 2020 and used OT during the labor process. The multiple linear regression method was used to screen the modeling variables from the electronic medical records, and combined with the variables such as fetal heart rate and uterine contraction frequency extracted by fetal heart rate, the OT drip speed prediction model was established based on XGBoost algorithm. At the same time, compared with Logistic regression model and traditional decision tree, the data set was divided as training set and test set in a ratio of 8:2, and the prediction performance of the model was evaluated in terms of accuracy, precision, recall rate and F1 score. Results: A total of 1 005 OT regulation records were included in 124 parturients with VBAC. The performance of XGBoost model was the best, with the under -50% discount cross-validation, accuracy, precision,recall and F1 value of the model being 0.82, 0.84, 0.80 and 0.82. The order of variable importance showed that the duration of uterine contraction, uterine pressure, fetal heart rate, frequency of uterine contraction and interval time of previous cesarean section were variables with great contributions to modeling. Conclusion: In this study,an OT regulation model is constructed based on XGBoost. It has the advantages of fast training speed, high accuracy and strong extrapolation, so it is of positive significance to assist the midwifery in the decision-making of OT injection.