基于经食道超声心动图超声数据的心脏病分类

发布时间：2018-03-17 09:40

本文选题：经食道超声心动图　切入点：超声数据　出处：《计算机应用》2017年S1期 　论文类型：期刊论文

【摘要】：为了研究经食道超声心动图(TEE)超声数据与心脏病类别之间的关系,提出一种以决策树(DT)分类器作为基分类器的自适应提升(AdaBoost)分类预测模型(DT_AdaBoost)。该模型首先对训练集中的每个样本赋予一个相同的权重表示样本被选中的概率,然后有放回地选取样本组成训练子集训练DT分类器,如果该分类器的分类准确率大于50%,则计算此DT分类器的重要性,并更新样本权重,最后在新的样本分布下再次进行抽样训练。依此类推,可得到多个权重不同的DT分类器,把所有DT分类器按重要性叠加(boost)起来,即可得到最终的强分类器。仿真结果表明,以DT分类器作为基分类器的Ada Boost方法诊断准确率相对稳定在96.88%,高于以支持向量机(SVM)作为基分类器的94.70%、以K最近邻(KNN)作为基分类器的94.65%以及以朴素贝叶斯(Naive Bayes)作为基分类器的96.04%,并且较单一算法的分类器性能提高。
[Abstract]:To study the relationship between transesophageal echocardiography (TEE) echocardiographic data and the type of heart disease, An adaptive lifting AdaBoost-based classification prediction model based on decision tree tree classifier (DTT) is proposed. In this model, each sample in the training set is first given an equal weight to indicate the probability of the sample being selected. Then a training subset is selected back to form a training DT classifier. If the classification accuracy of the classifier is greater than 50, the importance of the DT classifier is calculated, and the weight of the sample is updated. Finally, sampling training is carried out again under the new sample distribution, and so on, several DT classifiers with different weights can be obtained, all DT classifiers can be superimposed according to their importance, and the final strong classifiers can be obtained. The simulation results show that, The diagnostic accuracy of Ada Boost with DT classifier as base classifier is relatively stable at 96.88, which is higher than that with support vector machine (SVM) as base classifier (94.70), K nearest neighbor (KNN) as base classifier (94.65%) and naive Bayes (naive) as classifier. The base classifier is 96.04, and the performance of the single algorithm is improved.
【作者单位】：中国科学院成都计算机应用技术研究所;中国科学院大学;
【基金】：中国科学院西部之光人才培养计划项目(2013)
【分类号】：R540.45;TP311.13

【相似文献】