关于胎儿先天性心脏病优化诊断仿真研究

Simulation Study on Optimal Diagnosis of Fetal Congenital Heart Disease

对胎儿先天性心脏病问题的研究,能够有效提升对胎儿先天性心脏病的诊断质量。对先天性心脏病的优化诊断,需要将胎儿先天性心脏病数据样本特征变换到一个更高维的特征空间,对数据分类函数进行优化。传统方法得到胎儿先天性心脏病数据特征值间的相关性,提取数据特征,但忽略了获取数据分类函数以及对其的优化,导致诊断质量偏低。提出基于改进随机森林数据挖掘的胎儿先天性心脏病诊断方法。构建不同数量的数据样本分类回归树,计算出每个数据样本的特征,构建随机森林特征分类器,将数据样本特征低维空间的输入变量变换到一个更高维的特征空间,对数据的分类函数进行优化,以其优化的结果为依据进行胎儿先天性心脏病诊断。仿真证明,所提方法数据挖掘精度高,不仅可提高诊断的正确率,在对胎儿先天性心脏病的诊断中具有积极的使用价值。

A diagnosis method of fetal congenital heart disease based on improved random forest data mining is proposed. Firstly, different amounts of classification and regression tree of data samples are built, and characteristic of each data sample is calculated. Secondly, the classifier of random forest feature is constructed to transform input variables of data sample characteristic in low dimensional space to a higher dimensional feature space. Then, the classification function of data is optimized to diagnose fetal congenital heart disease based on optimization results. Simulation proves that the proposed method has high accuracy of data mining, which can not only improve the diag- nostic accuracy, but also have positive value in the diagnosis of fetal congenital heart disease.