SEER数据库的结直肠癌预后因素探讨及预后模型构建.pdfVIP

中华医学图书情报杂志2017年11月第26卷第11期摇 摇 ChinJ Med Libr Inf Sci,Vol.26 No.11摇 November,2017 ·7 · DOI:10.3969/ j.issn.1671-3982.2017.11.002 ·研究与探讨 · 基于SEER数据库的结直肠癌预后因素探讨及预后模型构建 辛世超 ,赵玉虹1 1,2 [摘要]分别使用logistic逐步回归法、贝叶斯模型平均法和 LASSO 回归进行特征变量筛选,分析美国SEER数据库的预后数 据,探讨影响结直肠癌预后的相关因素,并应用人工神经网络分类算法构建预后模型,指导结直肠癌预后评价。 结果证明,贝 叶斯模型平均法结合人工神经网络的混合算法所构建的预后模型准确率最高。 [关键词]结直肠癌;预后模型;特征选择;logistic逐步回归法;LASSO 回归;贝叶斯模型平均法 [中图分类号]G35;R730.7摇 摇 摇 摇 [文献标志码]A摇 摇 摇 摇 摇 [文章编号]1671-3982(2017)11-0007-07 Prognostic factors of colorectal cancer and establishment of colorectal cancer prognosis model 1 1,2 XIN Shi-chao ,ZHAO Yu-hong (1.China Medical University Medical informaticsSchool,Shenyang 110122,Liaoning Province,China;2. Affilia鄄 ted Shengjing Hospital of China Medical University,Shenyang 110004,Liaoning Province,China) Corresponding author:ZHAO Yu-hong [Abstract]The factors influencing the prognosis of colorectal cancer were studied after its characteristic variables were screenedby stepwiselogisticregressionanalysis,Bayesianmodelaveraginganalysis,andLASSOregressiona鄄 nalysis respectively. A model of colorectal cancer prognosis was established according to the artificial neural net鄄 work classification algorithmfortheassessment of colorectal cancer. Thehighest accuracywasdetectedinthemodel of colorectal cancer prognosis established by Bayesian model averaging analysis combined with artificial neural net鄄 work classification algorithm. [Keywords]Colorectalcancer;Prognosismodel;Selectionof characteristics;Logisticregressionanalysis;LASSO regression analysis;Bayesian model averaging analysis 摇 摇 结直肠癌包括结肠癌和直肠癌,是胃肠道中常 的最新统计显示,结直肠癌患者 5 年生存率仅为 见的恶性肿瘤。 在2015年中国癌症统计和2017 年 64.5%,中国结直肠癌患者5 年生存率比美国和欧 美国癌症统计中,结直肠癌的发病率和死亡率在所 [3] 洲更低 。 因此,建立结直肠癌预后模型,对制定