BayesCTDesign

BayesCTDesign provides Bayesian design and simulation capabilities in R for two-arm randomized trials, enabling incorporation of historical control data and power estimation across Gaussian, Poisson, Bernoulli, Weibull, Lognormal, and piecewise exponential outcomes.

Key Features:

• Bayesian framework: Integrates prior information, including historical control data, with current trial data within a Bayesian inferential framework.
• Integration of Historical Control Data: Incorporates historical control data as prior information to inform analyses of two-arm randomized trials.
• Simulation functions: Provides historic_sim() for scenarios with historical controls and simple_sim() for scenarios without, to simulate trial characteristics under user-defined scenarios.
• Outcome flexibility: Supports Gaussian, Poisson, Bernoulli, Weibull, Lognormal, and Piecewise Exponential (pwe) outcome models.
• Power estimation via simulation: Estimates power for two-sided hypothesis tests at a user-defined alpha by simulation, using 95% credible intervals for the treatment effect to assess rejection of the null.

Scientific Applications:

• Clinical trial design: Planning and evaluating two-arm randomized clinical trials with or without available historical controls.
• Randomized controlled trials: Assessing trial characteristics and operating characteristics for two-arm RCTs under varied assumptions and outcome models.
• Applied research across domains: Applying Bayesian simulation-based trial design to medical studies and ecological research that require Gaussian, count, binary, survival, or lognormal outcome modeling.

Methodology:

Operates within a Bayesian framework using prior (historical) information when available; implements simulation-based power estimation where each replication compares a 95% credible interval for the treatment effect to a null value to determine rejection, and provides the simulation functions historic_sim() and simple_sim() for data-generating scenarios.