基于Stacking模型的早产预测方法  

Preterm birth prediction framework under Stacking model

作　　者：马金龙 史晓月 杜丽佳 王胜普 杨志芬 MA Jinlong;SHI Xiaoyue;DU Lijia;WANG Shengpu;YANG Zhifen(School of Information Science and Engineering,Hebei University of Science and Technology,Shijiazhuang,Hebei 050018,China;Obstetrical Department,The Fourth Hospital of Hebei Medical University,Shijiazhuang,Hebei 050035,China)

机构地区：[1]河北科技大学信息科学与工程学院,河北石家庄050018 [2]河北医科大学第四医院产科,河北石家庄050035

出　　处：《河北工业科技》2025年第2期111-119,181,共10页Hebei Journal of Industrial Science and Technology

基　　金：河北省自然科学基金(H2022206212,H2022206600);河北省医学科学研究课题计划(20210715,20230775,20240817)。

摘　　要：为了解决传统机器学习模型在早产预测时综合性能不足的问题,提出一种基于Stacking模型的早产预测方法。首先,在数据预处理阶段,采用欠采样技术平衡正、负样本分布,并通过数据标准化消除变量间的数值差异;其次,通过分析特征之间的相关性和特征重要性分数,进行特征选择;然后,在Stacking模型构建时,通过分析机器学习算法预测结果间的皮尔逊相关系数,调整基分类器的类型和数量;最后,利用多种评价指标对基于Stacking模型的早产预测方法进行全面评估,并将其与现有方法对比分析,验证该方法的有效性。结果表明:所提方法在ROC曲线下面积(area under the curve,AUC)、准确率(Accuracy)、F 1值和召回率(Recall)方面,分别达到了0.9219、0.9229、0.9164和0.8585,均优于搭建Stacking模型所用的11个单一模型的最佳表现,且整体性能优于现有研究方法。所提方法能够高效识别孕早期的早产高风险人群,为早产的提前干预提供有力支持。To address the issue of insufficient overall performance of traditional machine learning algorithms in preterm birth prediction,an innovative preterm birth prediction method based on Stacking model was proposed.Firstly,during the data preprocessing stage,an under-sampling technique was applied to balance the distribution of positive and negative samples,and numerical differences between variables were eliminated through data standardization.Secondly,feature selection was carried out by carefully analyzing the correlations between features and assessing their importance scores.Then,in the construction of the Stacking model,the Pearson correlation coefficient was calculated among the prediction results of different machine learning algorithms,and this analysis was used to adjust both the type and number of base classifiers.Finally,a comprehensive evaluation of the preterm birth prediction method based on the Stacking model was conducted using multiple evaluation indicators,and compared and analyzed with existing methods to verify the effectiveness of the method.The results show that the proposed method achieves remarkable performance,with scores of 0.9219 in AUC,0.9229 in Accuracy,0.9164 in F 1 score,and 0.8585 in Recall.These results significantly outperform the best performances of the 11 individual models used to build the Stacking model,and the overall performance is better than the existing research methods.The proposed method can effectively identify high-risk individuals for preterm birth in early pregnancy,providing strong support for early intervention in early pregnanly.

关 键 词：人工智能其他学科 机器学习 集成学习 神经网络模型 早产预测 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]