基于变点检验的卵巢癌早期筛查法  

Screening Early Based on Changepoint Test for Ovarian Cancer

作　　者：刘媚[1] 张晓冉[2,3] 房祥忠[3] 

机构地区：[1]宁夏师范学院数学与计算机科学学院,固原756000 [2]燕山大学理学院统计学系,秦皇岛066004 [3]北京大学数学学院,北京100871

出　　处：《应用概率统计》2013年第6期593-606,共14页Chinese Journal of Applied Probability and Statistics

基　　金：国家自然科学基金(11361046);北京大学人民医院项目(首发2011-4022-02);河北省秦皇岛市科学技术研究与发展计划项目(201101A411)资助

摘　　要：文章对个体肿瘤标志物血清CA125的纵向检测数据建立变点模型,给出了模型参数最大似然估计的计算公式,构造出不依赖于模型参数的卵巢癌早期筛查的似然比检验统计量,并用随机模拟方法给出了部分情况下似然比检验统计量的临界值.文章最后讨论了似然比方法的功效和稳健性.In the paper a change-point model of ca125 longitudinal data for single subject was established. The formulae of the maximum likelihood estimate for model parameters were presented. A likelihood ratio test statistic for early screening of ovarian cancer was proposed. The critical values of likelihood ratio test statistic was calculated by Monte Carlo method in some cases. At last the power and robustness of the likelihood ratio test were discussed.

关 键 词：卵巢癌 纵向数据 变点模型 似然比检验 

分 类 号：O212.1[理学—概率论与数理统计]