节点文献
贝叶斯统计在高准确度筛检试剂评价中的应用
Application Bayesian statistics to evaluation of screening assays with high accuracy
【Author】 LIU Pei YANG Hai-tao~2,QIANG Lai-ying~3,JIN Hui~1,XIAO Shan~1.1.Department of Epidemiology and Biostatistics, School of Public Health,Southeast University,Nanjing 210009,China;2.Jiangsu Center for Disease Control and Prevention,Nangjing 210009,China;3.National Center for AIDS/STD Control and Prevontion,Chinese Center for Disease Control and Prevention,Beijing 100050,China
【机构】 东南大学公共卫生学院流行病与卫生统计学系; 江苏省疾病预防控制中心性病与艾滋病防制科; 中国疾病预防控制中心艾滋病预防控制中心;
【摘要】 目的针对经典统计的局限性,探讨贝叶斯统计在高准确度筛检试剂评价中的应用。方法通过分析艾滋病病毒抗体筛检试剂特征引入贝叶斯统计方法 ;通过贝叶斯统计与经典统计比较说明使用贝叶斯统计分析高准确度筛检试剂实验结果的合理性。结果当筛检实验的灵敏度、特异度等于100%时,贝叶斯统计可解决经典统计无法进行区间估计的难题;当灵敏度、特异度近于100%时,贝叶斯估计的平均值小于经典统计;当大样本时,无论先验分布取值如何,两种方法的结果接近。结论对高准确度筛检试剂评价,贝叶斯统计较经典统计结果合理。
【Abstract】 Objective In order to solve the limit of classic statistictics,application Bayesian statistics to evaluation of screening assays with high accurance.Methods Bayesian statistics methods were given by analyzing the screening assays for antibody to HIV.The rationality of using Bayesian statistics to evaluation of screening assays with high accurance were explained by comparing Bayesian statistics with classic statistics.Results When the sensitivity,specificity of screening test are equal to 100%,Bayesian statistics can overcome classic statistic’s difficulty of solving interval estimation.When the sensitivity and specificity are near to 100%,the average estimated by Bayesian statistics is smaller than that by classic statistics. When sample size is large enough,the results of Bayesian statistics are similar to classic statistics whatever prior distribution is chosen.Conclusions Bayesian statistics is prefer to classic statistics for evaluation of screening assays with high accurance.
- 【会议录名称】 华东地区第十次流行病学学术会议暨华东地区流行病学学术会议20周年庆典论文汇编
- 【会议名称】华东地区第十次流行病学学术会议暨华东地区流行病学学术会议20周年庆典
- 【会议时间】2010-08-23
- 【会议地点】中国安徽合肥
- 【分类号】R115
- 【主办单位】安徽省预防医学会、江西省预防医学会、江苏省预防医学会、浙江省预防医学会、山东省预防医学会、福建省预防医学会、上海市预防医学会