方法验证数据集的应用探讨  

Discussion on the application of method validation data set

作　　者：段丽 韩璐 杜颖 耿颖[1] 李娜[1] 许卉[2] 谭德讲[1] DUAN Li;HAN Lu;DU Ying;GENG Ying;LI Na;XU Hui;TAN De-jiang(NMPA Key lab for Quality Research and Evaluation of Chermical Drug,National Institutes for Food and Drug Control,Beijing 102629,China;Yantai University,Yantai 264003,China)

机构地区：[1]中国食品药品检定研究院国家药品监督管理局化学药品质量研究与评价重点实验室,北京102629 [2]烟台大学,烟台264003

出　　处：《药物分析杂志》2022年第6期972-978,共7页Chinese Journal of Pharmaceutical Analysis

基　　金：国家药品标准制修订课题(2022Y09);中检院质量安全与能力建设项目。

摘　　要：目的:充分挖掘方法验证的数据集,为方法后期的广泛应用提供更多参考依据。方法:利用严谨设计的方法验证所得数据集,通过对数据集中各方差分量的拆分或组合,构建日常实践中方法应用所需的评价依据或标准。结果:通过验证实验数据集的不同方差分量的拆分和组合,可形成3个应用评价指标或标准:(1)方法操作模式的变异度(FV);(2)报告值的区分阈值(RVDT);(3)实验室能力评定标准差(σ)。除此之外,还可利用验证数据集中影响因素的限制性最大似然(REML)方差分量值来辨识影响方法的主要因素,为实验操作的改进提供预判方向。结论:在方法验证过程中,使用科学严谨的验证实验设计所得的数据集,不仅可保证方法自身性能和方法检测能力的可靠评价,还可推导出更多有重要实践意义的应用评价标准,并对方法特性有更全面的了解。本文所论述指标,不仅适合于生物活性检测方法,也适合于理化方法。Objective: To validate the data set by the comprehensive mining method provides more reference for the wide application of the method in later stage. Methods: A method validation dataset was obtained by a rigorously designed method verification experiment, and the evaluation basis or standard required for the application of the method in daily practice was constructed by splitting or combining the different components of each party in the dataset. Results: By disassembling and combining the different variance components of the validation experimental dataset, the following three application evaluation indicators or criteria could be formed:(1) the format variability of method(FV);(2) the difference threshold of reportable value(RVDT);(3) and the standard deviation for proficiency assessment of the laboratory(σ). In addition, the REML variance component values of the influencing factors in the validation database could be used to determine the main factors influencing the method andprovide a prejudgment direction for the improvement of experimental operations. Conclusion: In the process of method validation, the dataset obtained by utilizing scientific and rigorous verification experimental design can not only guarantee the reliable evaluation of the method’s own performance and method detection ability, but also derive more application evaluation criteria with important practical significance and a comprehensive understanding of the method characteristics. The indexes discussed are not only suitable for bioassays, but also forphysio-chemical methods.

关 键 词：方法验证数据集 方法应用特性 操作模式变异度 报告值区分阈值 能力评定标准差 REML方差分量 

分 类 号：R917[医药卫生—药物分析学]