基于支持向量机的乳腺癌预后状态预测和疗效评估  

The prediction of prognosis and evaluation of curative effect for breast cancer based on support vector machine

作　　者：袁前飞[1] 蔡从中[1] 肖汉光[1] 刘兴华[1] 孔春阳[2] 

机构地区：[1]重庆大学应用物理系 [2]重庆师范大学物理学与信息技术学院,重庆400047

出　　处：《北京生物医学工程》2007年第4期372-376,共5页Beijing Biomedical Engineering

基　　金：重庆大学与新加坡国立大学国际联合科研项目(ARF-151-000-014-112);重庆市自然科学基金(CSTC;2006BB5240);重庆大学基础及应用基础研究基金(71341103)资助

摘　　要：乳腺癌是危害妇女健康的主要恶性肿瘤。目前基因与疾病关系的研究取得了一系列的成果,使得利用乳腺癌患者的基因信息来预测预后状态和评估治疗效果成为了可能。支持向量机(support vector machine,SVM)分类方法在实际二类分类问题的应用中显示出良好的学习和泛化能力,已被广泛地应用于诸多研究领域。本文采用支持向量机SVM、K-近邻法(K-nearest neighbor,K-NN)、概率神经网络(probabilistic neural network,PNN)、决策树(decision tree,DT)分类器,结合乳腺癌患者基因数据来预测患者的预后状态和评估治疗效果。结果表明:当使用高斯径向基核函数时,SVM通过5次交叉验证的最佳平均分类准确率达到了88.44%,优于K-NN(81.69%)、PNN(80.68%)和DT(71.19%)等分类器,表明该方法有望成为一种有效、实用的乳腺癌预后状态预测和治疗效果客观评价的工具。Breast cancer is mainly malignant tumour of endangering woman health. The investigation of the relationship between gene and disease has been achieved a series of outcomes, which afford a warranty for utilizing the genome information of breast cancer patient to predict the prognosis and evaluate the curative effect. The Support Vector Machine （SVM） has shown its excellent learning and generalization ability in the practice problems of binary classification, and has been extensively employed in many fields. In this paper, based on the gene data of breast cancer patient, the SVM, K-Nearest Neighbor （K-NN）, Probabilistic Neural Network （PNN） and Decision Tree （DT） were applied to predict the prognosis and evaluate the curative effect of breast cancer patient. The best overall accuracy reached 88.44% via SVM with RBF kernel function by using 5 - fold cross validation, which is superior to those of other classifiers based on K-NN （81.69%）, PNN （80.68%） and DT （71.19%）. This study suggests that SVM is capable of being used as a potential application and efficiency tool for predicting the prognosis and objective evaluating the curative effect of breast cancer.

关 键 词：支持向量机 乳腺癌 预后 评估 

分 类 号：R318.04[医药卫生—生物医学工程]