BayesianMCPMod 1.3.2 (14-May-2026)

• Included Firth’s penalized regression model for binary endpoints in case of separation.

BayesianMCPMod 1.3.1 (25-Feb-2026)

• Fixed a newly introduced bug that would occur if the R package future.apply was not installed.
• Added flexibility to bootstrapped credible bands in plot.modelFits().

BayesianMCPMod 1.3.0 (23-Feb-2026)

• Fixed a bug that would occur when predicting from the beta model shape outside of the original dose range.
• Fixed a bug in which the MED assessment could not be performed when specifying a negative direction of beneficial effect and an evidence level other than 0.5.
• Added functions and vignettes for the binary endpoint case.
• Added functionality to assessDesign() to provide custom simulated data and custom model estimates enabling complex data simulation and analysis methods.
• Added argument to assessDesign() for number of bootstrap samples in case evidence_level is provided.
• Added functionality to plot.modelFits() to plot effect sizes.
• Added calls to set.seed() in vignette's code blocks to facilitate individual code block reproducibility.

BayesianMCPMod 1.2.0 (28-Aug-2025)

• Fixed a bug in performBayesianMCPMod() where the model significance status from the MCP step was sometimes not correctly assigned to the fitted model in the Mod step.
• Fixed a bug in print.modelFit() where sometimes the coefficients for the fitted model shapes were not printed correctly.
• Fixed a bug in getMED() where quantile and evidence level could sometimes not be matched due to floating-point precision issues when using bootstrapped quantiles.
• Changed functions getPosterior(), getCritProb(), and getContr() to accept a covariance matrix instead of a standard deviation vector as argument.
• Added support for none-zero off-diagonal covariance matrices in the MCP step.
• Added bootstrapped differences to getBootstrapSamples().
• Added average MED identification rate as attribute to assessDesign() output.
• Made the future.apply package optional.
• Re-worked vignettes and improved the output of print functions.

BayesianMCPMod 1.1.0 (07-Mar-2025)

• Fixed a bug in plot.modelFits() that would plot credible bands based on incorrectly selected bootstrapped quantiles
• Added getMED(), a function to assess the minimally efficacious dose (MED) and integrated getMED() into assessDesign() and performBayesianMCPMod
• Added parallel processing using the future framework
• Modified the handling of the fit of an average model: Now, getModelFits() has an argument to fit an average model and this will be carried forward for all subsequent functions
• Re-introduced getBootstrapSamples(), a separate function for bootstrapping samples from the posterior distributions of the dose levels
• Adapted the vignettes to new features

BayesianMCPMod 1.0.2 (06-Feb-2025)

• Addition of new vignette comparing frequentist and Bayesian MCPMod using vague priors
• Extension of getPosterior to allow the input of a fully populated variance-covariance matrix
• Added the non-monotonic model shapes beta and quadratic
• New argument in assessDesign() to skip the Mod part of Bayesian MCPMod
• Additional tests

BayesianMCPMod 1.0.1 (03-Apr-2024)

Re-submission of the ‘BayesianMCPMod’ package
Removed a test that occasionally failed on the fedora CRAN test system
Fixed a bug that would return wrong bootstrapped quantiles in getBootstrappedQuantiles()
Added getBootstrapSamples(), a separate function for bootstrapping samples

Initial release of the 'BayesianMCPMod' package

BayesianMCPMod 1.0.0 (31-Dec-2023)

• Initial release of the 'BayesianMCPMod' package
• Special thanks to Jana Gierse, Bjoern Bornkamp, Chen Yao, Marius Thoma & Mitchell Thomann for their review and valuable comments
• Thanks to Kevin Kunzmann for R infrastructure support and to Frank Fleischer for methodological support