网络Meta分析研究进展系列(十九):基于外部证据确立贝叶斯网络Meta分析异质性的先验信息

Advances in methodology of network meta-analysis(19):Informative prior based on external evidence for heterogeneity estimation using bayesian network meta-analysis

本文围绕医学研究证据整合领域异质性问题,介绍了异质性先验分布类型、使用外部证据确立异质性先验信息的普适的经验,进一步聚焦于异质性的方差-协方差矩阵讲述了异质性估计常用的四种模型。期望帮助读者掌握如何使用外部经验数据作为异质性方差的先验信息,得到更稳健的网络Meta分析估计。

Focusing on the topic of heterogeneity in the field of medical research evidence integration,this paper introduces the types of prior distribution of heterogeneity and the helpful experience of using external evidence to establish informative prior of heterogeneity,and further focuses on the variance-covariance matrix of heterogeneity to describe four commonly used models of heterogeneity estimation.It is expected to help readers master how to use external empirical data as a priori information of heterogeneous variance to obtain more robust estimations of network meta analysis.