Release Notes MonolixSuite 2019R2

Monolix

Assessement tab:

“use linearization” button availability was not properly updated when fixing parameters
The initial parameters list was not updated upon model change
Error parameters were not updated if the error model type was changed

Data tab:

Columns and headers were shifted when a data file had been loaded prior to editing a header type
A name defined as a covariate or a regressor in the data set header can only be used in the model as a covariate or a regressor, respectively
For continuous observation models, unreachable values and censoring limits in the data set are now generating a warning
The content of data tab was left disabled when creating a new project after the fit had been modified and not saved
Columns tagging experienced performance issues with moderate datasets
Outdated results & plots tabs are now hidden after data edition
The number of observation types was not properly updated after modifying data tagging
The names of subject-occasions were wrong in case of redundant columns OCCASION and EVENT ID
The user’s order of duplicated observations given at same time are now kept

Statistical model tab:

Group names of transformed categorical covariates are now properly checked for validity to avoid errors
The formula for the observation model was hidden when the observation model was changed
In some cases, warning messages were prevented from being shown in “Statistical Model & Tasks” tab upon project loading
GUI remained disabled after a bad load and a fit edit submission
Changing the default number of iterations for the task “Standard errors” could cause an error

Check initial estimates tab:

In some cases several results were possible fot auto-init and chosen randomly
The reference curve was not updated when changing the output or discretization grid
“fix all” were not fixing all betas if there are more than 2 modalities in categorical covariate
Default initial values of omega’s for Mtt parameters from library models are now set to 1 instead of 2

Structural model tab:

some hidden files of the model libraries (identified by a leading dot) could appear in the list

Results tab:

Stopping criterion in optimization of error models has been modified
Individual parameter tests fixed if there is IOV and missing variability at last levels

[export] Sublabels were not visible from SVG exports in some viewers (Inkscape, Powerpoint)
[preferences] Comma-separated values for stroke dasharrays were not working anymore
[stratify] Changing composite groups color sometimes crashed
[covariate viewer] In the case of several categorical covariates, the y-axis limits for cat vs cat barplots were wrong
[individual fits] In case of IOV, empty subject-occasions are now hidden if they are not surrounded by non-empty subject-occasion
[scatter residuals] Exporting and reloading settings with different bin criteria resulted in an error to generate the plot
[vpc continuous] The prediction correction option is now also applied on observations
[cov vs indivParams] NaNs are properly managed for boxplots computation
[residuals distribution] The plot was not properly generated when empirical PDF length was <2
[tte] Continuous time grid for empirical survival curves was restored

[Load] Setting checks is more robust
[Load] Estimation was not possible following a load error and setting a new dataset
[Load] GUI was disabled following a load error and setting a new dataset
[Settings] When ‘Set default’ is clicked in project settings, the export path is now updated also for the current session
[Save] After “Save as” with results, some result directories were missing
[Save] Changing the error model and observation distribution caused a loading error in some cases

[Load] Consistency of settings for NCA additional tasks (partialAUCTime, ajdR2AcceptanceCriteria, extrapAUCAcceptanceCriteria, spanAcceptanceCriteria) with the saved project is now checked when reloading old results
[Load] Additional points included for lambda_z estimation are saved with exact values
[Plots] After each run of a task, the associated plots are now rebuilt
[Plots] Plots for NCA parameters with NaNs had wrong display
[Data] It was not possible to select urine mode
[Data] Previous data & tasks settings are now kept when possible after changing the data
[Data] Fixed conflicts between some header names and CA model
[NCA] Saving with a new name a project with results which has not run in the same session does not crash anymore
[NCA] Missing CDISC names (partial AUC dependent, ajdR2AcceptanceCriteria, extrapAUCAcceptanceCriteria, spanAcceptanceCriteria, Ctau) have been added
[NCA] Additional point at dose time for steady-state extravascular or infusion is now min over the dosing interval, not over all observations
[NCA] Discrepancies in calculation of AUClast in case of steady state data have been fixed
[NCA] For infusion adminsutrations, points prior to the end of the infusion are now excluded from the lambda_z when using a main rule
[CA] Individual fits plot available after model edition

[PKanalix] Arguments dataType, administrationType and timeValuesPerID were not set to lowercase and thus were only recognized in lowercase
[PKanalix] Display of estimated individual parameters has been fixed in case of NaN values
[PKanalix] Covariates have been added in fields of getCAIndividualParameters and getNCAIndividualParameters
[Monolix] The header of a covariate was changed with setData if the same name was used as regressor in the model, this situation is no longer possible
[Monolix] An error in setPopulationParameterEstimationSettings() function caused by the setting simulatedannealingiterations has been fixed => this setting has been hidden
[Commons] The output structure of getObservationInformation() function has been fixed when there are several observation models, and one of which has subject-occasions without observation
[Commons] Mapping information between data and model has been added to getObservationInformation() & getProjectInformation() functions outputs
[Commons] An error in setPreferences() function caused by the setting headeraliases has been fixed
[Commons] Launching initializeLixoftConnectors(path = ‘pathToMonolixSuite’) generates the lixoft.ini file if not present