posologyr v1.2.7

All vignettes are moved to articles
The updates to poso_simu_pop() in v1.2.5 introduced several issues and have been reverted
The new function poso_replace_et() enables updating a model with events from a new rxode2 event table, while accounting for and interpolating any covariates or inter-occasion variability

posologyr v1.2.6

Use undirected quotes when quoting in the DESCRIPTION (as requested by CRAN)
CRAN test times have been reduced by using an environment variable to identify the development machine, which now determines whether less critical tests are executed (as requested by CRAN)

The route of administration (i.e. the compartment in which the drug is to be administered) can now be specified in poso_time_cmin(), poso_dose_conc(), poso_dose_auc() and poso_inter_cmin().
poso_simu_pop() provides an rxode2 model using the simulated ETA and the input dataset, with interpolation of covariates, to make plotting easier

The README illustrates a simple example of dose adaptation
vignette("route_of_administration") shows how to select a route of administration for optimal dosing
vignette("population_models") describes the structure of prior population models written as model functions which can be parsed by rxode2 and used by posologyr
vignette("posologyr_user_defined_models") is renamed vignette("classic_posologyr_models")
Examples use rxode2 model functions

Fix a bug where poso_estim_map(), poso_estim_sir() and poso_simu_pop() failed for models featuring a single parameter with IIV.

Add ability to use rxode2 ui models for the poso_* functions. Once the model has been parsed by rxode2() with this package the model$posologyr gives the list needed for poso_* functions

Fix a bug in poso_dose_conc(), poso_dose_auc() and poso_inter_cmin() where the returned estimate of the target value to be optimized against was always equal to zero.

The documentation for poso_time_cmin(), poso_dose_conc(), and poso_dose_auc() now explicitly states the consequences of setting tdm to TRUE: which parameters are required, which parameters are ignored, and which parameters behave differently.
The functions poso_time_cmin(), poso_dose_conc(), and poso_dose_auc() now return a warning if any of the input parameters are ignored.
Fix incorrect information regarding the duration of the AUC in the documentation of poso_dose_auc()

Relax the requirements of the NONMEM comparison test for time-varying covariates to account for computational differences observed with the alternative BLAS ATLAS on CRAN.

Add a reference to Kang et al. (2012) doi:10.4196/kjpp.2012.16.2.97 in the DESCRIPTION (as requested by CRAN)
Fix messages to the console in the internal function posologyr() (as requested by CRAN)
Fix assignment to parent environment in dose optim functions, using parent.frame() (as requested by CRAN)

poso_estim_map(), poso_estim_sir() and poso_estim_mcmc() can now estimate individual PK profiles for multiple endpoints models (eg. PK-PD, parent-metabolite, blood-CSF…), using a different residual error model for each endpoint.
poso_time_cmin(), poso_dose_conc(), poso_dose_auc() and poso_inter_cmin() now allow you to select the end point of interest for which you want to optimise, provided it is defined in the model.

vignette("a_priori_dosing") illustrates a priori dose selection
vignette("a_posteriori_dosing") illustrates a posteriori dose selection, using TDM data
vignette("auc_based_dosing") shows how to select an optimal dose for a given target AUC using data from TDM
vignette("multiple_endpoints") introduces the new multiple endpoints feature

poso_time_cmin() can now estimate time needed to reach a selected trough concentration (Cmin) using the data from TDM directly
poso_dose_conc() can now estimate an optimal dose to reach a target concentration following the events from TDM
poso_dose_auc() can now estimate an optimal dose to reach a target auc following the events from TDM

posologyr() is now an internal function, all exported functions take patient data and a prior model as input parameters
The adaptive MAP forecasting option is removed

poso_estim_map() provides an rxode2 model using MAP-EBE and the input dataset, with interpolation of covariates, to make plotting easier

RxODE import is updated to rxode2
All tests are updated to take into account the internalization of the posologyr() function

poso_time_cmin(), poso_dose_auc(), poso_dose_conc(), and poso_inter_cmin() no longer fail for models with IOV

poso_estim_sir() estimates the posterior distribution of individual parameters by Sequential Importance Resampling (SIR). It is roughly 25 times faster than poso_estim_mcmc() for 1000 samples.
poso_estim_map() allows the estimation of the individual parameters by adaptive MAP forecasting (cf. doi: 10.1007/s11095-020-02908-7) with adapt=TRUE.
poso_simu_pop(), poso_estim_map(), and poso_estim_sir() now support models with both inter-individual (IIV) and inter-occasion variability (IOV).
MASS:mvrnorm is replaced by mvtnorm::rmvnorm for multivariate normal distributions.
Input validation is added to all exported functions.
poso_estim_map() now uses method=“L-BFGS-B” in optim for better convergence of the algorithm.
poso_inter_cmin() now uses method=“L-BFGS-B” in optim for better convergence of the algorithm.
poso_dose_conc() is the new name of poso_dose_ctime().
Issues #5 and #6 are fixed: poso_time_cmin(), poso_dose_auc(), poso_dose_conc(), and poso_inter_cmin() now work with prior and posterior distributions of ETA, and not only with point estimates (such as the MAP).
A new nocb parameter is added to posologyr(). The interpolation method for time-varying covariates can be either last observation carried forward (locf, the RxODE default), or next observation carried backward (nocb, the NONMEM default).
vignette("uncertainty_estimates") is removed.
The built-in models are removed.

poso_time_cmin(), poso_dose_ctime(), and poso_dose_auc() now work for multiple dose regimen.
poso_inter_cmin() allows the optimization of the inter-dose interval for multiple dose regimen.
vignette("case_study_vancomycin") illustrates AUC-based optimal dosing, multiple dose regimen, and continuous intravenous infusion.

First public release.