基于Stacking模型融合的胎儿健康状态智能评估  

Intelligent Evaluation of Fetal Health Status Based on Stacking Ensemble Model

作　　者：郝婧宇 陈奕[2] 吴水才[1] HAO Jingyu;CHEN Yi;WU Shuicai(Faculty of Environmental and Life,Beijing University of Technology,Beijing 100124,China;Department of Obstetrics and Gynecology,Beijing Obstetrics and Gynecology Hospital,Capital Medical University,Beijing 100124,China)

机构地区：[1]北京工业大学环境与生命学部,北京100124 [2]首都医科大学附属北京妇产医院妇产科,北京100124

出　　处：《中国医疗设备》2022年第7期19-25,共7页China Medical Devices

基　　金：国家自然科学基金(71661167001)。

摘　　要：目的研究机器学习算法评估妊娠期间胎儿在子宫内的状态,提出一种基于Stacking模型融合的胎儿宫内状态智能评估新方法。方法在特征选择阶段,运用极端梯度提升树与热力图对公开的胎心数据集分析,选择出最优特征子集。在分类阶段,运用一种两层Stacking模型融合新方法对胎儿进行评估,第一层集合5种强机器学习模型来训练,第二层采用Logistics回归模型。结果运用胎心数据测试集来验证,分类准确率达0.950,受试者曲线下面积达0.980。结论基于Stacking模型融合的新方法可辅助临床医师对胎儿宫内健康状态进行诊断。Objective To study a machine learning algorithm for evaluating fetal state in utero during pregnancy,and to propose a new intelligent evaluation method for fetal state in utero based on Stacking ensemble model.Methods In the feature selection stage,XGBoost and HeatMap were used to analyze the public fetal heart data set,and the optimal feature subset was selected.At the classification stage,the fetus was evaluated using a new method fused with a two-layer Stacking model,the first layer was trained with five strong machine learning models,and the second layer was trained with a Logistics regression model.Results Using the fetal heart data test set,ACC and AUC were 0.950 and 0.980 respectively.Conclusion A new method based on Stacking ensemble model can assist clinicians in the diagnosis of fetal intrauterine health status.

关 键 词：胎心监护 极端梯度提升树算法 Stacking模型融合 机器学习算法 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]