基于正则化的胎心监护图智能分类算法研究

Research on an intelligent classification algorithm for fetal heart monitoring graphs based on regularization

目的:基于凸优化理论探讨智能评估胎儿在子宫中健康状态的方法,快速准确评估胎儿在宫内的健康状态。方法:在详细描述胎心监护数据参数含义的基础上,阐述基于数据的核范数正则化评估方法;采用波尔图大学公开的真实数据集,在Matlab软件环境下实施算法流程,计算该方法用于胎儿健康状态分类的准确性。结果:基于正则化的胎心监护图智能分类算法优化了数据量过小所导致的过拟合问题,准确率最高达92.14%,高于逻辑回归模型、卷积神经网络方法、RF+GBDT+AdaBoost+XGBoost融合模型以及最小二乘支持向量机的算法准确率。结论:本算法可以辅助临床医师对胎儿状态进行智能评估,提高临床决策的准确性。

Objective Based on the convex optimization theory,this paper explores the intelligent methods for evaluating the fetal health status in utero,so as to quickly and accurately evaluate fetal health status in utero.Methods On the basis of describing the parameters of fetal heart monitoring data in detail,the kernel norm regularization evaluation method based on data was elaborated.Using the real data set published by the University of Porto,the algorithm flow was implemented in the Matlab software and the accuracy of the method for fetal health classification was calculated.Results The regularized intelligent classification algorithm for fetal heart monitoring graphs optimized the overfitting problem caused by too small amount of data,with the highest accuracy of 92.14%,which was higher than that of LR,CNN,RF+GBDT+AdaBoost+XGBoost model and least squares support vector machine.Conclusion This algorithm can assist clinicians to evaluate the fetal status intelligently and improve the accuracy of clinical decision-making.