The bonsaiforest2 package has been developed to facilitate Bayesian
shrinkage estimation of treatment effects in subgroups of randomized
clinical trials. It supports both one-way models, which fit a
separate Bayesian shrinkage model for each subgrouping variable, and
global models, which fit a single model including all subgroup
variables at once. For both types of models, treatment effect estimates
in subgroups are derived via standardization (G-computation). The
package supports continuous, binary, time-to-event (Cox), and count
outcomes. It leverages brms and Stan for model fitting, allowing for
a flexible choice of priors, including normal, regularized horseshoe,
and R2-D2.
A comprehensive description of the methodology implemented in this R package can be found in Wolbers et al. (2026).
The latest stable release of bonsaiforest2 can be installed from the
main branch on GitHub:
remotes::install_github("openpharma/bonsaiforest2")bonsaiforest2 depends on the brms
package and a R interface to Stan (e.g.,
cmdstanr)
This example demonstrates Bayesian shrinkage estimation of treatment
effects across subgroups using a global model with a regularized
horseshoe prior (with default hyperprior parameters from brms),
applied to the shrink_data dataset with three subgrouping variables.
library(bonsaiforest2)
# 1. Load the shrink_data package dataset
shrink_data <- bonsaiforest2::shrink_data
# 2. Fit global shrinkage model
fit_fixed <- run_brms_analysis(
data = shrink_data,
response_type = "continuous",
response_formula = y ~ trt,
unshrunk_terms_formula = ~ x1 + x2 + x3,
shrunk_predictive_formula = ~ 0 + trt:x1 + trt:x2 + trt:x3,
shrunk_predictive_prior = "horseshoe(1)",
chains = 2, iter = 1000, warmup = 500,
backend = "cmdstanr", refresh = 0
)# 3. Extract and visualize standardized treatment effects in subgroups
subgroup_effects <- summary_subgroup_effects(fit_fixed)
print(subgroup_effects$estimates)
#> # A tibble: 9 × 4
#> Subgroup Median CI_Lower CI_Upper
#> <chr> <dbl> <dbl> <dbl>
#> 1 x1: a 0.486 0.249 0.719
#> 2 x1: b 0.105 -0.243 0.436
#> 3 x2: a 0.388 0.114 0.629
#> 4 x2: b 0.421 0.155 0.713
#> 5 x2: c 0.267 -0.00707 0.508
#> 6 x3: a 0.329 0.0389 0.545
#> 7 x3: b 0.407 0.157 0.722
#> 8 x3: c 0.384 0.110 0.727
#> 9 x3: d 0.313 0.0216 0.544
plot(subgroup_effects)
Wolbers, Marcel, Miriam Pedrera Gómez, Alex Ocampo, and Isaac Gravestock. 2026. “Unified Implementation and Comparison of Bayesian Shrinkage Methods for Treatment Effect Estimation in Subgroups.” arXiv Preprint arXiv:2603.21967. https://arxiv.org/abs/2603.21967.