贝叶斯统计在高准确度筛检试剂评价中的应用

Application of Bayesian Statistics to Evaluation of Screening Assays with High Accuracy

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。方法引入贝叶斯统计方法分析艾滋病病毒抗体筛检试剂特征;通过贝叶斯统计与经典统计比较,阐述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。

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 is equal to 100%,Bayesian statistics can overcome classic statistic’s difficulty of solving interval estimation.When the sensitivity and specificity is 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.Conclusion Bayesian statistics is prefer to classic statistics for evaluation of screening assays with high accurance.