摘要
目的探讨无金标准诊断实验条件下,贝叶斯先验参数的确定方法,并比较不同方法的应用条件。方法根据贝叶斯共轭先验分布原理,对二项分布的共轭贝塔分布中的α、β两个先验参数的确定方法进行分析比较,编写SAS程序确定先验参数。结果在共轭先验分布的条件下,先验矩、分位数、众数与分位数三种方法确定的先验分布参数结果一致。结论在实际工作中,应根据已知条件和具体情况决定采用何种方法计算先验分布参数。
To study the methods of estimation of prior parameter and compare their application condition in the absence of the gold standard based on the Bayesian theory. Methods According to： the Bayesian conjugate prior distribution principle, the estimating methods of prior parameters （a、β） of the conjugate beta distribution Be （a、β） of the binomial distribution was given by writing the SAS programs. Results Under the condition of the conjugate prior distribution, the prior parameters computed by three methods were similar. Conclusion The different methods for estimation of prior parameter should be used according to known conditions and practical situation.
出处
《中国卫生统计》
CSCD
北大核心
2007年第3期232-234,共3页
Chinese Journal of Health Statistics
基金
国家自然科学基金资助项目(项目编号30471501)
关键词
贝叶斯统计
先验分布
先验参数
无金标准
诊断实验
Bayesian statistics
Prior distribution
Prior parameter
Absence of the gold standard
Diagnostic test
