Package index • BExTE

Data generation

sample_aggregate_binary_data(): Sample aggregate binary data based on the rate and sample size
sample_aggregate_normal_data(): Sample aggregate normal data based on the mean and variance
sample_log_odds_ratios(): Sample log odds ratios based on the number of participants and responders in each arm
compute_ORs(): Compute odds ratios based on the number of participants and responders in each arm
compute_log_odds_ratio_from_counts(): Compute the log odds ratio from count data
rate_from_drift_logOR(): Calculate the success rate in the target study arm from the drift on the log odds ratio scale
rate_from_drift_logRR(): Calculate the rate in the target study arm from the drift on the log relative risk scale
standard_error_log_odds_ratio(): Compute the standard error of the log odds ratio

Methods

BinomialPooling: BinomialPooling class
BinomialSeparate: BinomialSeparate class
GaussianConjugate: GaussianConjugate class
GaussianPDCCPP: GaussianPDCCPP class
GaussianPooling: GaussianPooling class
RecurrentEventTargetData: Recurrent Event Target Data
GaussianSeparate: GaussianSeparate class
GaussianStaticBorrowing: GaussianStaticBorrowing class
TestThenPool: TestThenPool class
TestThenPoolDifference: TestThenPoolDifference class
TestThenPoolEquivalence: TestThenPoolEquivalence class
GaussianEmpiricalBayesPP: GaussianEmpiricalBayesPP class
GaussianGravestockEBPP: GaussianGravestockEBPP class
GaussianNPP: GaussianNPP class
GaussianNPP_KL: GaussianNPP_KL class
BinomialCPP: BinomialCPP Class
BinomialNPP: BinomialNPP class
BinomialGravestockEBPP: BinomialGravestockEBPP class
BinomialLatticePrior: BinomialLatticePrior class
BinomialNPP_KL: BinomialNPP_KL class
BinomialPDCCPP: BinomialPDCCPP class
BinomialRMP: BinomialRMP class
BinomialEgidiMixture: BinomialEgidiMixture class
BinomialCommensuratePowerPrior: BinomialCommensuratePowerPrior class
BinomialCommensuratePrior: BinomialCommensuratePrior class
GaussianCommensuratePowerPrior: GaussianCommensuratePowerPrior class
GaussianCommensuratePrior: GaussianCommensuratePrior class
GaussianEgidiMixture: GaussianEgidiMixture class
GaussianRMP_RBesT: GaussianRMP_RBesT class
Model_RBesT: Model_RBesT
GaussianPooling_RBesT: GaussianPooling_RBesT
GaussianSeparate_RBesT: GaussianSeparate_RBesT
MCMCModel: MCMCModel class
Model: Model Class
BinomialPValueBasedPP: GaussianEmpiricalBayesPP class
GaussianPValueBasedPP: GaussianPValueBasedPP class
BinomialConjugate: BinomialConjugate class

Binomial quadrature

Exact posteriors of a difference in response rates for the binomial borrowing models, computed on a grid instead of sampled

binomial_power_prior_posterior(): Posterior of the treatment effect under the binomial power prior
binomial_effect_grid(): Grid of treatment effects covering a binomial posterior
beta_quadrature_nodes(): Quantile midpoints of a Beta distribution
beta_sd(): Standard deviation of a Beta distribution
grid_posterior(): Summarise a density tabulated on a grid
grid_posterior_cdf(): Distribution function of a grid posterior
grid_posterior_pdf(): Density of a grid posterior
grid_posterior_quantile(): Quantiles of a grid posterior
grid_posterior_sample(): Draw from a grid posterior

Vectorised inference

Conjugate updates and frequentist tests evaluated across all replicates of a scenario at once

normal_mixture_posterior(): Conjugate update of a normal mixture prior across replicates
analyse_in_replicate_chunks(): Run a vectorised analysis of the replicates in chunks
combine_replicate_results(): Concatenate the results of consecutive chunks of replicates
fit_egidi_mixture(): Fit the Egidi, Pauli and Torelli empirical mixture prior
normal_mixture_summary(): Mean, standard deviation and quantiles of a normal mixture, per replicate
normal_mixture_elir_ess(): ELIR effective sample size of a normal mixture, per replicate
npp_prior_mixture(): Discretise the normalised power prior as a normal mixture
calibrate_npp_kl(): Calibrate the normalised power prior by a KL criterion
summary_t_test_p_value(): Two-sample t-test from summary statistics, across replicates

Plots

method_style(): Fixed shape and hue for a method
method_key(): Resolve a method to its configuration key
method_shape_map(): Shapes for a set of methods, keyed by method
method_hue_map(): Base hues for a set of methods, keyed by method
method_hue_ramp(): A light-to-dark ramp through a method's hue
method_parameter_colors(): Colours for the parameter values of one method
method_parameter_color_map(): Colours for the parameter values present in one method's rows
method_parameter_positions(): Where each parameter label sits on its method's ramp
style_config_object(): Resolve a configuration object the plot code keeps in the global environment
style_blend(): Blend a colour towards another
style_label_numbers(): Numbers a parameter label assigns to its parameters
style_numeric_ranges(): The numeric parameter ranges a method's configuration declares
style_tuple_key(): A tuple of parameter values as a lookup key
plot_metric_vs_scenario(): Plot methods operating characteristics
forest_plot(): Generate a forest plot
estimate_bayesian_ocs(): Simulation Bayesian OCs
set_size(): Function to set figure dimensions
nominal_tie_breaks(): Axis breaks that always show the nominal type-I error
has_monte_carlo_uncertainty(): Whether a pair of bound columns carries Monte Carlo uncertainty
plot_baseline_success_proba_vs_xvar(): Add a reference probability of success to a metric-vs-x-variable plot
bayesian_metric_vs_parameters(): Function to plot metric vs parameters
bayesian_metric_vs_sample_size(): Function to plot metric vs sample size
bayesian_ocs_plots(): Plot methods operating characteristics
frequentist_power_at_equivalent_tie(): Compute the frequentist power at equivalent tie
frequentist_power_at_nominal_tie(): Compute the frequentist power at the nominal type I error rate
plot_metric_vs_drift(): Function to generate a metric vs drift plot
plot_metric_vs_parameters(): Function to plot metric vs parameters
plot_metric_vs_sample_size(): Function to plot metric vs sample size
power_vs_tie(): Function to generate a power vs tie plot
posterior_parameters_plots(): Plot frequentist methods operating characteristics
power_vs_tie_plots(): Plot frequentist methods operating characteristics
forest_plot_methods_comparison(): Plot methods comparison
forest_plot_bayesian(): Generate a forest plot for bayesian metrics
forest_plot_methods_comparison_bayesian(): Plot methods comparison
plot_metrics_vs_ess(): Plot methods operating characteristics
plot_posterior_parameters_vs_drift(): Function to plot posterior parameters vs drift
plot_success_proba_vs_drift(): Function to generate a metric vs drift plot
plot_success_proba_vs_scenario(): Plot methods operating characteristics
plot_empirical_bayes_hyperparameters_vs_drift(): Plot Empirical Bayes Hyperparameters vs Drift
empirical_bayes_parameters_plots(): Plot the posterior parameters
bayesian_operating_characteristics_vs_tie_plots(): Plot methods operating characteristics
forest_plot_sweet_spot(): Generate a forest plot
forest_plot_sweet_spots_comparison(): Plot methods comparison
operating_characteristics_vs_tie_plots(): Plot methods' operating characteristics
operating_characteristic_vs_tie(): Plot an operating characteristic against the type I error rate
plot_metric_vs_drift_methods(): Function to generate a metric vs drift plot
plot_sweet_spot_width_vs_scenario(): Plot methods operating characteristics
plot_metric_vs_sample_size_methods(): Function to plot metric vs sample size
plot_metric_vs_scenario_methods(): Plot methods operating characteristics
scale_by_separate(): Express a metric as a ratio to the separate analysis

Operating characteristics

average_power(): Calculate the average power
average_tie(): Calculate the average type 1 error
sweet_spot(): Calculate Sweet Spots for Multiple Metrics
sweet_spot_determination(): Determine Sweet Spot Bounds and Width
upper_bound_proba_FP_MC(): Calculate the upper bound probability of false positive
compute_freq_power(): Compute the frequentist power
preposterior_proba_FP_MC(): Calculate the probability of false positive
preposterior_proba_TP_MC(): Calculate the probability of true positive
prior_moment_ess(): Calculate the prior moment-based Effective Sample Size (ESS) for a Bayesian model.
prior_precision_ess(): Calculate the prior precision-based Effective Sample Size (ESS) for a Bayesian model.
gaussian_mix_moment_ess(): Calculate the moment-based Effective Sample Size (ESS) for a Gaussian mixture.
gaussian_mix_precision_ess(): Calculate the precision-based Effective Sample Size (ESS) for a Gaussian mixture.
compute_freq_power_pooling(): Compute the frequentist power
prior_ess_elir(): Calculate the priorEffective Sample Size (ESS) based on ELIR for a Bayesian model.
normal_reference_ess(): Effective sample sizes of a posterior summarised against a normal reference
adaptive_power_prior_type_I_error(): Type I error of the adaptive power prior, computed exactly
log_grid_interpolation(): Interpolate an expensive function of a positive scalar
interval_score(): Interval score of a credible interval
compute_power_with_tie_ci(): Compute the frequentist power at an estimated type I error

Monte Carlo uncertainty

Propagating simulation error into the comparison with the frequentist benchmark at equivalent type I error

binomial_replicate_count(): Recover the number of Monte Carlo replicates behind a binomial estimate
sample_binomial_proportion(): Draw from the posterior of a binomial proportion
mover_difference_ci(): Confidence interval for a difference, by variance estimate recovery
add_power_difference_columns(): Replace success probabilities by power differences
flag_power_differences(): Flag power gains and power losses against the comparator

Simulation

compute_control_drift_range(): Compute the control drift range for a given source treatment effect.
compute_drift_range(): Compute the drift range for a given simulation and case study configuration.
compute_source_denominator_range(): Compute the source denominator range for a given source data, simulation configuration, and case study configuration.
compute_time_to_event_ranges(): Compute the ranges of the time-to-event design axes.
frequentist_ocs_scenario_simulation(): Simulate a Scenario
bayesian_ocs_scenario_simulation(): Simulate a Scenario
simulation_analysis(): Perform Simulation Analysis
simulation_bayesian_ocs(): Run simulations based on the given environment
simulation_frequentist_ocs(): Run simulations based on the given environment
simulation_scenarios(): Generate simulation scenarios based on the configuration directory.
run_simulation_env(): Run a full simulation for one environment
run_progress_tracker(): Track how many scenarios of a run have been simulated.
time_to_event_expected_events(): Expected number of observed events in each arm of the target trial
time_to_event_delayed_log_hr(): Post-delay log hazard ratio that gives a target Cox estimand
is_primary_time_to_event_design(): The design a time-to-event case study is primarily simulated under
time_to_event_axes_off_primary(): How many time-to-event design axes are off their primary value
time_to_event_sensitivity_reference(): The point at which the time-to-event sensitivity designs are simulated

Tables

table_methods_comparison(): Methods comparison in a table
posterior_parameters_tables(): Generate multiple posterior parameter tables
power_vs_tie_tables(): Generate power vs tie tables for multiple conditions
table_power_vs_tie(): Generate a power vs tie table
table_bayesian_ocs(): Generate tables for methods operating characteristics
table_metric_vs_drift(): Generate a metric vs drift table
table_metric_vs_parameters(): Plot metric vs parameters
table_metric_vs_sample_size(): Plot metric vs sample size
table_posterior_parameters_vs_drift(): Generate a table of posterior parameters vs drift
tables_metric_vs_scenario(): Generate tables for methods operating characteristics
table_empirical_bayes_hyperparameters_vs_drift(): Generate a table of hyperparameters estimated using Empirical Bayes vs drift
table_empirical_bayes_hyperparameters_vs_scenario(): This function tables for hyperparameters updated using Empirical Bayes across different case studies, methods, and sample sizes.
table_bayesian_metrics(): Generate tables for Bayesian metrics
table_case_study_summary(): Summary of the clinical case studies (table S3)
table_drift_ranges_and_sample_sizes(): Drift ranges and target sample sizes for each case study (table S1)

Paper replication

The manifest of the paper’s figures and tables, and what a results directory needs to reproduce them

paper_manifest(): The paper figure and table manifest
paper_manifest_ids(): Ids of every paper item in the manifest
paper_manifest_entry(): Look up one manifest entry by id
paper_sample_size_per_arm(): Resolve a sample size factor to a target sample size per arm
paper_replication_requirements(): Minimal simulation config for a set of paper figures
paper_replication_coverage(): Check a results frame against a set of paper figures
paper_run_config(): The simulation config a results directory was produced from
paper_config_shortfalls(): How a results directory falls short of the paper's fidelity
export_paper_outputs(): Produce the paper's figures and tables

Data

TargetData: TargetData class
TargetDataFactory: TargetDataFactory class
SourceData: SourceData class
BinaryTargetData: BinaryTargetData class
ObservedSourceData: ObservedSourceData class
ObservedTargetData: ObservedTargetData class
ContinuousTargetData: ContinuousTargetData class
TimeToEventTargetData: Time To Event Target Data

Design priors

DesignPrior: DesignPrior Class
AnalysisPriorDesignPrior: AnalysisPriorDesignPrior class
SourcePosteriorDesignPrior: SourcePosteriorDesignPrior class
UnitInformationDesignPrior: UnitInformationDesignPrior class

Caching

Inspecting and clearing the compiled-model and analysis caches

inference_cache_size(): Number of analyses held in the cache
inference_cache_reset(): Empty the analysis cache
generate_replicates(): Generate the replicates of a scenario, reusing them across methods
generation_cache_key(): Key of a set of generated replicates
npp_kl_calibration_cache_size(): Number of calibrations held in the cache
npp_kl_calibration_cache_reset(): Empty the KL calibration cache
clear_stan_draws(): Remove the Stan draws of one model across every process
clear_stan_model_cache(): Remove the compiled Stan models

Utilities

check_colnames(): Check columns in a dataframe
combine_parameters(): Combine parameters for a given method.
append_parameters_str(): Append a parameter string to a figure or table filename
convert_params_to_str(): Function to convert parameters in dataframe to string
hellinger_distance(): Compute the Hellinger distance between two normal distributions.
important_drift_values(): Get important drift values for a given source treatment effect and case study configuration.
make_labels_from_parameters(): Function to make labels from parameters. Return a label formatted in Tex, for example "$\xi_\gamma$ = 0.5, $\sigma_\gamma$ = 0.1"
negative_binomial_regression(): Estimate rate and standard error
findCalibrationParameter(): Find Calibration Parameter
get_parameters(): Function to return a dataframe of parameters from json strings
remove_columns_from_df(): Remove columns from a dataframe
generate_comparison_table(): Generate a comparison table
read_function_code(): Print a function's source code
format_simulation_output_table(): Format simulation output as a printable table
format_case_study_config(): Format a case study configuration as a printable table