论文部分内容阅读
目的分析支持向量机(SVM)对结肠腺瘤高级别上皮内瘤变的发生预测效果。方法随机抽取2008—2010年北京大学人民医院消化科行结肠息肉高频电凝切除术患者157例,收集每例患者临床及内镜相关的12个指标,利用SVM及Logistic回归分析两种方法分别对结肠腺瘤高级别上皮内瘤变的发生进行预测。在应用SVM预测部分,首先随机抽取50例作为训练集,利用LIBSVM-2.88软件建立预测模型,再从剩余病例中随机抽取3组,作为3个测试集,应用训练集所得的模型对3个测试集分别进行预测,得出平均预测正确率及特异度、敏感度。在应用Logistic回归分析部分,全部157例患者均纳入Logistic回归模型进行分析。最后对SVM及Logistic回归模型两种方法的预测结果进行比较。结果 SVM预测分类结果 3个测试集平均预测正确率为(92.6±3.3)%,平均敏感度为(80.6±17.3)%,平均特异度为(94.8±0.6)%。Logistic回归模型预测正确率为90.4%。结论应用SVM建立的预测模型在小样本的基础上对结肠腺瘤发生高级别上皮内瘤变获得了较好的预测效果。
Objective To analyze the predictive effect of support vector machine (SVM) on high-grade intraepithelial neoplasia of colonic adenoma. Methods A total of 157 patients with colonic polyp hyperplasia resected from Peking University People’s Hospital from 2008 to 2010 were randomly selected. Twelve indexes of clinical and endoscopic were collected in each patient. SVM and Logistic regression were used to analyze the two methods Predict the occurrence of high-grade intraepithelial neoplasia of colon adenoma. In the application of SVM prediction, 50 samples were randomly selected as training set, LIBSVM-2.88 software was used to establish the prediction model, then 3 random sets were selected from the remaining cases as three test sets. The training set was used to test three tests Set separately predict, come to the average prediction accuracy and specificity, sensitivity. In the application of Logistic regression analysis, all 157 patients were included in the Logistic regression model for analysis. Finally, the prediction results of two methods of SVM and Logistic regression model are compared. Results The average prediction accuracy rate of the three test sets was (92.6 ± 3.3)%. The average sensitivity was (80.6 ± 17.3)% and the average specificity was (94.8 ± 0.6)%. Logistic regression model predicts the correct rate of 90.4%. Conclusion The prediction model established by SVM has good predictive value for high-grade intraepithelial neoplasia of colon adenoma based on the small sample.