metavidence - No-Code 'shiny' Application for Meta-Analysis
A 'shiny' application that guides researchers through
meta-analysis in three steps: load the data, choose the
parameters and review the results. Covers pairwise
meta-analysis of binary and continuous outcomes, pre-calculated
effect sizes, single-arm proportions and means, frequentist
network meta-analysis and diagnostic test accuracy, producing
forest plots, funnel plots, leave-one-out analyses, subgroup
analyses, meta-regression and publication-ready exports.
Pairwise and single-arm models use the methods implemented in
'meta' and 'metafor'; network meta-analysis follows Rucker
(2012) <doi:10.1002/jrsm.1058> as implemented in 'netmeta';
diagnostic accuracy uses the bivariate model of Reitsma et al.
(2005) <doi:10.1016/j.jclinepi.2005.02.022>, fitted with 'lme4'
as a bivariate binomial generalized linear mixed model (Chu and
Cole (2006) <doi:10.1016/j.jclinepi.2006.06.011>) following the
Cochrane Handbook for Systematic Reviews of Diagnostic Test
Accuracy. The development and validation of the application are
described in Costa Filho et al. (2026)
<doi:10.64898/2026.11.17.26368328>. The application runs
locally and no data leaves the user's computer.