beastt - v0.0.3

What's Changed

• sweetspotplot() added to make it easier to graphically summarise simulation results
• The following simulation functions added to make it easier to simulate binary and time to event data:
  • bootstrap_cov() Bootstrap Covariate Data
  • calccondbinary() and calccondweibull() to convert a vector of drift and treatment effects for the conditional to the marginal scale
  • inv_logit()
  • sim_accrual() to simulate participant accrual times
  • simweibph() to simulate event times for each participant from a Weibull proportional hazards regression model
  • simpwconst_haz() to simulate event times for each individual from a piecewise constant hazard model
  • calccondweibull() to calculate conditional drift and treatment effect for time-to-event outcome models
  • calcstudyduration() to calculate the analysis time based on a target number of events and/or target follow-up time
• trimps() and rescaleps() functions added to trim and re-scale the propensity score object
• propscrcloud() function added to visualise propensity scores
  

In addition we have added template simulation code into inst/templates. These additions should make it significantly easier to not just do IPW BDB, but to simulate how the addition of historical data will affect the operating characteristics of your trial.

- R
Published by statasaurus about 1 year ago

beastt - Time to Event

What's Changed

The biggest addition to this release was the inclusion of time to event modelling. This expands the scope of {beastt} to cover beta, normal and Weibull power priors. This also means that we have started to include {rstan} into the package and no longer just do closed form models.

There was also a bug fix for all three calc_power_prior_* functions, changing the weights from 1 to a vector of 1s for cases when external data are read in directly rather instead of a prop_scr_obj

- R
Published by statasaurus over 1 year ago

beastt - First CRAN Release

This release sets the foundation for Bayesian dynamic borrowing with covariate adjustment via inverse probability weighting for simulations and data analyses in clinical trials. It makes it easy to use propensity score methods to balance covariate distributions between external and internal data.

At the moment the package covers the normal case with known and unknown standard deviation as well as the beta-binomial case.

- R
Published by statasaurus almost 2 years ago