条件相关假设下条件似然方法在无“金标准”诊断试验准确性评价中的应用

Application of conditional likelihood method in the evaluating of diagnostic test without a gold standard under conditional correlation assumption

目的旨在介绍条件相关假设下,使用经验似然法估计无“金标准”诊断试验准确性评价相关参数的方法。方法基于已有的Hui-Walter模型构建条件简化模型,即在条件相关假设下,通过构建条件似然函数,得到诊断试验评价相关参数的估计。结果该方法在条件相关假设下通过构建条件似然函数,得到包括当前样本患病率、灵敏度及特异度等相关参数可靠且稳健的估计值。结论该方法简化了无“金标准”诊断试验分析的条件,扩展了无“金标准”诊断试验评价的应用范围,是一种新的行之有效的无“金标准”诊断试验准确性评价的方法。

Objective To introduce the method of using empirical likelihood method to estimate the relevant parameters of accuracy evaluation of diagnostic test without a gold standard under conditional correlation hypothesis.Methods Based on the existing Hui-Walter model,the conditional simplified model is constructed,that is,under the conditional correlation hypothesis,the estimation of the relevant parameters of the diagnostic test evaluation is obtained by constructing the conditional likelihood function.Results By constructing the conditional likelihood function under the conditional correlation hypothesis,the method obtains reliable and robust estimates of related parameters,including the prevalence rate,sensitivity and specificity of the current sample.Conclusion This method simplifies the conditions for the analysis of the diagnostic test without a gold standard,expands the application range of evaluation of diagnostic test without a gold standard,and is a new and effective method to evaluate the accuracy of diagnostic test without a gold standard.