基于支持向量机模型预测结直肠癌根治术预后风险

发布时间：2018-02-26 03:34

本文关键词： 结直肠癌风险预测支持向量机　出处：《重庆医科大学学报》2017年02期 　论文类型：期刊论文

【摘要】：目的:基于支持向量机(support vector machine,SVM)预测结直肠癌患者术后风险。方法:回顾性分析2011年1月至2013年12月行根治性手术治疗的199例结直肠癌患者临床资料,按照是否无病生存(disease free survival,DFS)分为预后好、差2组。采用SAS9.2进行基本统计分析、logistic回归分析,运用R3.2.4软件建立SVM预测模型、受试者工作曲线(receiver operating characteristic curve,ROC)及曲线下面积(area under the curve,AUC)评价模型预测效果。结果:差异性分析显示,总蛋白、白蛋白、阳性淋巴率、组织类型、p53、N分期、放化疗等因素在预后组间存在统计学差异(P0.05);多因素逐步logistic回归结果显示,N2分期(OR=13.5,95%CI=3.85~47.5)、p53阳性(OR=0.314,95%CI=0.125~0.790)可能是影响预后的危险因素。线性核、多项式核、sigmoid核及RBF(径向基函数)核SVM模型的预测准确率分别为69.70%、66.70%、69.70%、74.24%。对所有预测模型性能对比显示,SVM模型预测性能高于logistic回归模型,其中RBF-SVM模型效果最好,预测准确率为74.24%,特异度为0.714,灵敏度为0.765,AUC值为0.708(95%CI=0.551~0.866)。结论:N2分期、p53阳性可能是影响预后结局的危险因素,应当予以重视,减少预后风险。
[Abstract]:Objective: to predict the postoperative risk of colorectal cancer patients based on support vector machine support vector machine. Methods: the clinical data of 199 patients with colorectal cancer undergoing radical surgery from January 2011 to December 2013 were retrospectively analyzed. Patients with disease-free free survival were divided into two groups with good prognosis and poor prognosis. SAS9.2 was used to carry out basic statistical analysis and logistic regression analysis. R3.2.4 software was used to establish SVM prediction model. The predictive effect of the model was evaluated by receiver operating characteristic curvec and area under curve under the curve.Results: total protein, albumin, positive lymphoid rate, tissue type and p53 N stage were found by differential analysis. The multivariate stepwise logistic regression analysis showed that the N _ 2 staging was 13.5C ~ 95 CI 3.85 / 47.5% p53 positive and 0.314% 95% CI 0.125% 0.125% -0.790) which may be the risk factors of prognosis. Linear nucleus, linear nucleus, linear nucleus, and so on were found to be the main risk factors for the prognosis of the patients with different prognosis (P < 0.05), and the results of stepwise logistic regression showed that the p53 positive rate was 0.31495% 0.125% 0.125% (P < 0.05). The prediction accuracy of polynomial kernel sigmoid kernel and RBF-based kernel SVM model is 69.70 and 69.700.Compared with all the prediction models, it is shown that the prediction performance of RBF-SVM model is better than that of logistic regression model. The predictive accuracy was 74.24, the specificity was 0.714, and the sensitivity was 0.708 / 95CI0.551C / 0.8660.Conclusion the positive expression of p53 in N _ 2 staging may be a risk factor affecting the outcome of prognosis, which should be paid more attention to and reduce the risk of prognosis.
【作者单位】：重庆医科大学公共卫生与管理学院卫生统计与信息管理教研室;
【基金】：重庆市自然科学基金资助项目(编号:cstc2013jcyj A0068)
【分类号】：R735.34

【参考文献】