Logistic回归分析在前列腺癌诊断中的应用价值  

Application of Logistic Regression Analysis in the Diagnosis of Prostate Cancer

作　　者：陈媚聪[1] 朱贤胜[1] 王莎莎[1] 程琦[1] 王泓[1] 贺冬莲[1] 

机构地区：[1]广州军区广州总医院超声科,广东广州510010

出　　处：《华南国防医学杂志》2013年第4期244-247,共4页Military Medical Journal of South China

摘　　要：目的探讨logistic回归模型对前列腺癌(prostate cancer,PCa)的诊断价值。方法回顾性分析120例患者的151个经病理证实的前列腺结节的临床资料及超声特征,建立预测前列腺癌的Logistic回归模型,绘制模型的受试者工作特征(receiver operating characteristic,ROC)曲线并计算曲线下面积。结果进入Logistic回归模型的变量为前列腺特异性抗原(prostate specific antigen,PSA)、游离前列腺特异性抗原(free prostate specific antigen,FPSA)及FPSA/TPSA比值、血流分级、年龄。Logistic回归模型的曲线下面积为0.926,该指标在0.375时获得最大的敏感性(90.0%)和特异性(84.2%)。结论Logistic回归模型的建立对前列腺良恶性结节的鉴别诊断有较高的应用价值。Objective To evaluate the diagnostic value of logistic regression model for prostate cancer. Methods The ultrasonographic characteristics and clinical data of 151 solid prostate masses in 120 patients diagnosed by pathology were analyzed retrospectively. The logistic regression model to predict prostate cancer was established, the receiver operat- ing characteristic （ROC） curve was drawn and the areas under curve （AUC） were calculated. Results Four features,i, e. prostate-specific antigen （PSA）, ratio of total PSA and free PSA （FPSA/TPSA）, grade of blood flow and age were finally included in the logistic regression model. The AUC was 0. 926. When the AUC was 0. 375, the maximum sensitivity （90. 0%） and specificity （84. 2//oo） could be obtained. Conclusion The logistic regression model is helpful to identify be- nign and malignant prostate masses.

关 键 词：前列腺癌 超声检查 LOGISTIC回归 受试者工作特征曲线 

分 类 号：R737.25[医药卫生—肿瘤]