Overview

The bcts package provides functions to simulate, calibrate, and evaluate Bayesian non-inferiority (NI) and superiority trials with binary outcomes. It implements conjugate Beta–Binomial models, with support for:

• Flat priors (no external evidence),
• Power priors incorporating external control data.

The package is designed for both methodological research and practical trial planning, offering tools to:

• Simulate trial operating characteristics (Type-I error, power, posterior probabilities),
• Calibrate posterior thresholds (γ) to target frequentist error rates (α),
• Evaluate borrowing strength from external data,
• Visualize calibration traces, prior/posterior distributions, and effective sample sizes.

A companion Shiny app is available to explore designs interactively.

---

Installation

Development version (from GitHub)

# install.packages("remotes")
remotes::install_github("smartdata-analysis-and-statistics/bcts")

Launch the Shiny App

After installation, launch the interactive Shiny app by running:

bcts::run_bcts_app()

This will open a browser window where you can:

• Simulate randomized and single-arm Bayesian trials,
• Calibrate posterior decision thresholds,
• Estimate Type-I error and power under different priors,
• Explore visualizations of posterior distributions and calibration traces.

The app supports trial design using Bayesian Beta–Binomial models and is intended to assist both applied users and methodologists.