论文部分内容阅读
目的:探讨血清肿瘤标志物检测结合支持向量机(SVM)模型在口腔鳞癌诊断中的价值。方法:应用酶联免疫吸附法及时间分辨荧光分析法分别测定163例口腔鳞癌患者和160例健康志愿者血清神经元特异性烯醇化酶(NSE)、癌抗原242(CA242)、癌抗原19-9(CA199)、癌胚抗原(CEA)、组织多肽抗原(TPA)、癌抗原72-4(CA724)、细胞角蛋白19片段(CA211)和甲胎蛋白(AFP)的含量;采用SVM进行数据分析,并建立诊断模型,以留一法进行交叉验证,评价SVM模型在口腔鳞癌诊断中的价值。结果:根据肿瘤标志物的不同组合,经过SVM建模、训练和验证,筛选出以CA211、CA199、TPA、CA724和NSE这5个肿瘤标志物组合为最佳的SVM模型用于口腔鳞癌诊断,其诊断的准确率、特异度、敏感度和阳性预测值分别为88.54%、93.13%、84.05%和92.57%。结论:本研究筛选出的5个最佳血清肿瘤标志物组合结合SVM模型,对口腔鳞癌的诊断具有较高价值。
Objective: To investigate the value of serum tumor marker detection combined with support vector machine (SVM) model in the diagnosis of oral squamous cell carcinoma. Methods: The serum levels of neuron specific enolase (NSE), cancer antigen 242 (CA242), cancer antigen 19 (CA199), carcinoembryonic antigen (CEA), tissue polypeptide antigen (TPA), cancer antigen 72-4 (CA724), cytokeratin 19 fragment (CA211) and alphafetoprotein (AFP) Data analysis, and to establish a diagnostic model to leave a method for cross-validation, evaluation of SVM model in the diagnosis of oral squamous cell carcinoma. Results: According to the different combinations of tumor markers, SVM model was screened for the best diagnosis of oral squamous cell carcinoma by SVM modeling, training and verification, and the best combination of these five tumor markers CA211, CA199, TPA, CA724 and NSE The diagnostic accuracy, specificity, sensitivity and positive predictive value were 88.54%, 93.13%, 84.05% and 92.57% respectively. Conclusion: The five best serum tumor markers screened in this study combined with SVM model have high value for the diagnosis of oral squamous cell carcinoma.