支持向量机原理及其在医学分类中的应用

The Principle of Support Vector Machine and its Application in Medical Classification

目的介绍一种具有较高精度的分类模型——支持向量机在解决分类问题时的应用。方法以胃癌流行病学调查资料为例分别建立支持向量机、决策树、logistic回归模型,比较三种模型性能优劣。结果对于测试集样本SVM预测精度为99.052%,C5.0决策树预测精度为93.365%,logistic回归预测精度为90.995%,SVM具有良好的泛化能力。结论当传统统计分析条件不能得到满足或效果不佳时支持向量机能够达到良好的预测结果,在医学领域具有较好的应用前景。

Objective To introduced a high Precision classification model--Support Vector Machine and its application in medical classification. Methods Build Support Vector Machine, Decision Tree and logistic regression model using Epidemiological survey real data of gastric cancer to comparison of prediction accuracy between three models. Results For testing set the prediction accuracy of SVM is 99. 052%, C5.0 decision tree is 93. 365% and logistic regression is 90.995%, SVM has good generalization ability. Conclusion Support Vector Machine displays the advantage when conditions of classical statistical techniques could not be met or the predictive effect is bad, and will make a better facture of its apphcation in medical researches.