🔬 Network Meta-Analysis and Its Role in Evidence-Based Decision-Making

Master both Frequentist and Bayesian approaches to Network Meta-Analysis

📚 Tutorial • 🚀 Quick Start • 📊 Examples • 🛠️ Installation • 📖 Documentation

🎯 What is Network Meta-Analysis?

Network Meta-Analysis (NMA) is a powerful statistical technique that allows researchers to compare multiple treatments simultaneously, even when some treatments have never been directly compared in head-to-head trials. This tutorial provides a comprehensive guide to conducting NMA using both frequentist and Bayesian statistical frameworks.

🔥 Why This Tutorial?

📈 Dual Approach: Master both frequentist (netmeta) and Bayesian (multinma) methods
🎨 Rich Visualizations: Generate professional plots including evidence networks, forest plots, and rankograms
🔍 Model Diagnostics: Learn advanced techniques for heterogeneity assessment and consistency checking
📊 Real Data: Work with actual clinical trial datasets (Stowe2010 and parkinsons)
🎓 Educational: Step-by-step explanations suitable for beginners to advanced users

📚 Tutorial Overview

🔵 Frequentist Approach

Package: netmeta
Dataset: Stowe2010

✅ Evidence network visualization
✅ Forest plots for treatment effects
✅ Rankograms and SUCRA values
✅ Heterogeneity decomposition
✅ Consistency checking with net-splitting
✅ Network heat maps

🔴 Bayesian Approach

Package: multinma
Dataset: parkinsons

✅ Network setup with set_agd_arm()
✅ Fixed and Random Effects models
✅ DIC and residual deviance
✅ UME models for consistency
✅ Prior-posterior visualizations
✅ Treatment rankings and probabilities

🚀 Quick Start

Prerequisites

# Install required packages
install.packages(c("netmeta", "multinma", "ggplot2", "dplyr", "gridExtra"))

Run the Tutorial

# Clone the repository
git clone https://github.com/dgalgom/NMA-webinar-r.git
cd NMA-webinar-r

# Run the main tutorial
source("NMA_webinar.R")

📊 Examples

🌐 Evidence Networks

Frequentist Network	Bayesian Network

📈 Treatment Rankings

Rankograms	Cumulative Probabilities

🛠️ Installation

System Requirements

R: Version 4.0 or higher
RStudio: Recommended for best experience
Operating System: Windows, macOS, or Linux

Package Dependencies

# Core packages
library(netmeta)      # Frequentist NMA
library(multinma)     # Bayesian NMA
library(ggplot2)      # Plotting
library(dplyr)        # Data manipulation
library(gridExtra)    # Plot arrangements

# Additional packages (installed automatically)
# - meta, magic, MASS, coda, rjags, etc.

Docker Setup (Optional)

FROM rocker/tidyverse:latest

RUN R -e "install.packages(c('netmeta', 'multinma', 'gridExtra'))"

COPY . /home/rstudio/nma-tutorial
WORKDIR /home/rstudio/nma-tutorial

📖 Documentation

📋 File Structure

nma-tutorial-r/
├── 📄 README.md                 # This file
├── 📜 nma_tutorial.R            # Main tutorial script
├── 📊 data/                     # Sample datasets
├── 📈 outputs/                  # Generated plots
├── 📚 docs/                     # Additional documentation
│   ├── frequentist_guide.md     # Frequentist methods
│   ├── bayesian_guide.md        # Bayesian methods
│   └── interpretation_guide.md  # Results interpretation
├── 🧪 examples/                 # Additional examples
└── 🔧 utils/                    # Helper functions

🎯 Key Functions Overview

Frequentist Functions (netmeta)

Function	Purpose	Output
`netmeta()`	Conduct network meta-analysis	NMA object
`netgraph()`	Plot evidence network	Network visualization
`forest()`	Create forest plots	Treatment effects plot
`rankogram()`	Generate rankograms	Treatment ranking plot
`decomp.design()`	Assess heterogeneity	Heterogeneity breakdown
`netsplit()`	Check consistency	Consistency assessment
`netheat()`	Network heat map	Inconsistency visualization

Bayesian Functions (multinma)

Function	Purpose	Output
`set_agd_arm()`	Set up network data	Network object
`nma()`	Conduct Bayesian NMA	NMA model
`plot()`	Visualize networks/results	Various plots
`relative_effects()`	Calculate treatment effects	Effect estimates
`posterior_ranks()`	Treatment rankings	Ranking probabilities
`plot_prior_posterior()`	Compare distributions	Prior vs posterior

🎨 Generated Outputs

The tutorial generates 13 high-quality plots:

📊 Frequentist Outputs

evidence_network_freq.png - Network structure
forest_plot_freq.png - Treatment effects
rankogram_freq.png - Treatment rankings
decomp_design.png - Heterogeneity decomposition
netsplit_plot.png - Consistency checking
netheat_plot.png - Network heat map

📈 Bayesian Outputs

evidence_network_bayes.png - Network structure
resdev_contrib.png - Residual deviance
devdev_plot.png - Dev-dev plot
prior_posterior.png - Prior vs posterior
relative_effects.png - Treatment effects
treatment_ranking.png - Rankings
cumulative_ranking.png - Cumulative probabilities

🐛 Bug Reports

Found a bug? Please create an issue with:

Description: Clear description of the problem
Reproducible Example: Minimal code to reproduce the issue
Environment: R version, package versions, OS
Expected vs Actual: What you expected vs what happened

💡 Feature Requests

Have ideas for improvements? We'd love to hear them! Please include:

Use Case: Why this feature would be useful
Description: Detailed description of the proposed feature
Examples: How it would work in practice

📊 Use Cases

This tutorial is perfect for:

👩‍🔬 Researchers conducting systematic reviews and meta-analyses
👨‍⚕️ Clinicians evaluating treatment effectiveness
🎓 Students learning network meta-analysis methods
📊 Statisticians comparing frequentist and Bayesian approaches
🏥 Health Technology Assessment agencies

🔗 Resources

📚 Further Reading

🛠️ Related Tools

netmeta R package
multinma R package
R-INLA - Alternative Bayesian framework

📄 Citation

If you use this tutorial in your research, please cite:

@software{nma_tutorial_r,
  author = {Your Name},
  title = {Network Meta-Analysis in R: A Comprehensive Tutorial},
  url = {https://github.com/yourusername/nma-tutorial-r},
  year = {2025},
  version = {1.0.0}
}

🙋‍♂️ Support

Need help? Here's how to get support:

📖 Documentation: Check the docs folder
💬 Discussions: Use GitHub Discussions for questions
🐛 Issues: Report bugs via GitHub Issues
📧 Email: Contact daniel.gallardo@iconplc.com

🌟 Star History

Star History

🎉 Happy Analyzing!

Made with ❤️ by the ICON Clinical Research team

⬆️ Back to Top

Name		Name	Last commit message	Last commit date
Latest commit History 4 Commits
NMA_webinar.R		NMA_webinar.R
README.md		README.md

dgalgom/NMA-webinar-by-ICON-plc

Folders and files

Latest commit

History

Repository files navigation