BUGSnet

BUGSnet facilitates Bayesian network meta-analysis by automating BUGS code generation for integration with JAGS and implementing generalized linear model–based Bayesian NMAs to estimate treatment effects, assess heterogeneity and inconsistency, and produce diagnostic and ranking outputs.

Key Features:

• BUGS code generation and JAGS integration: Automates generation of BUGS code from inputs and executes Bayesian NMAs via JAGS (Just Another Gibbs Sampler).
• Bayesian generalized linear model estimation: Implements generalized linear models within a Bayesian framework to estimate treatment effects.
• Network description: Provides functions to describe evidence networks and produce network characteristic tables and network plots.
• Model estimation and diagnostics: Estimates models and assesses model fit and convergence using diagnostics such as traceplots and posterior mean deviance comparison plots.
• Heterogeneity and inconsistency assessment: Evaluates heterogeneity and inconsistency within network meta-analytic data.
• Analytical outputs and ranking: Generates league tables, league heat plots, SUCRA plots, rankograms, forest plots, leverage plots, and data plots.
• Guideline mapping: Maps package functions and outputs to checklist items within reporting and regulatory guidelines.
• Example replication: Demonstrated by recreating a Bayesian NMA from the NICE Decision Support Unit (NICE-DSU).

Scientific Applications:

• Comparative effectiveness research: Enables simultaneous comparison of multiple interventions in healthcare through Bayesian network meta-analysis.
• Evidence synthesis and decision support: Supports evidence-based decision-making, systematic review reporting, and guideline-informing analyses.

Methodology:

Automated generation of BUGS code; Bayesian estimation via JAGS (Just Another Gibbs Sampler) using generalized linear models; model estimation with assessment of model fit and convergence; evaluation of heterogeneity and inconsistency.