2017.12.05
2017年12月12日に久留米大学バイオ統計センター公開セミナーを開催します。
日時：2017年12月12日（火）16:00-17:00
場所：バイオ統計センター 講義室
講演者： Leslie Jayasekara
（Professor in Mathematics、University of Ruhuna,
Sri Lanka）
演 題：Bivariate meta-analysis when the within-study correlations
are unknown: A Novel Bayesian Approach
概 要：Bivariate meta-analysis can be used to combine two outcome
measures instead of only concerning one specific outcome
measure. We need to quantify the correlations between
outcomes across studies when fitting bivariate models.
The model requires an estimate of the correlations between
treatment effect for each study. These correlations are known
as within study correlations and rarely available in published
literature. This limits the bivariate approach in practice.
Therefore, we have to either approximate them or ignore
correlations between effect estimates within the same studies.
Some widely applicable ways in which this problem can be
resolved are discussed here. We also examine the use of the
Bayesian correlation method as a novel approach for dealing with
unknown within-study correlations in bivariate meta-analysis.
An application to a meta-analysis of treatments for acute stroke
data illustrates the use of the approximated correlation in
bivariate meta-analysis with correlated outcomes.
