Release Notes for Pumas 2.3.2


Bug fixes and minor improvements

  • Added a cor2cov method for PDMats such that it works on params with LKJCholesky priors.
  • Support fixed width string types (String3, etc) in read_pumas and Subject constructors because they may be be present when reading in tabular data.
  • Speed up DataFrame constructor for SimulatedObservations (output of simobs).

Pumas Utilities

Bug fixes and minor improvements

  • Fix a bug where apps did not work when using more than two vCPUs.



  • Statistical significance now uses Satterthwaite degrees of freedom
  • Support for additional replicated designs
  • Results include estimates of within-subject variability CV
  • Provide convenience functions for the sigma to geometric CV and in reverse.
  • Include functionality to conveniently simulate various reference-scaled confidence intervals and their critical boundary estimates.
  • Adapted the reference scaling from the ABEL framework to using the critical boundary re-specification according to latest regulatory guidance.
  • Reference scaling parameter changed from regulatory and auxiliary parameters to (theta - regulatory parameter) and cv_max for maximum within-subject variability for expanding the confidence interval.
  • Added the homogeneity argument to control the degrees of freedom for parallel designs and the model structure for replicated crossover designs.
  • Estimates for the geometric mean ratio and associated confidence intervals are reported in level scale rather than in percent form.
  • Added a assess_be to obtain the bioequivalent assessment from a study based on different and customizable criteria including pre-defined criteria for major regulatory guidances.

Bug fixes and minor improvements

  • Column mapping arguments now accept AbstractString in addition to Symbol arguments.
  • Reference argument removed in favor of alphabetical order of set letters, e.g. RT, RST, and ABCD.