Skip to content

Latest commit

 

History

History
129 lines (91 loc) · 4.76 KB

NEWS.md

File metadata and controls

129 lines (91 loc) · 4.76 KB
Raw

nlmixr2lib

Version 0.3.0

  • Added ability to choose style type when modifying models. Currently supported styles are: "camel" for variablesLikeThis, "snake" for variables_like_this, "dot" for variables.like.this and "blank" for variableslikethis. This can be selected with setCombineType().

  • With the new combination style, you can change how eta variables are constructed with the option(nlmixr2lib.etaCombineType="camel") or whatever you wish it to the variable style to be.

  • Added new model building framework for building models

    • PK model building functions

      • addTransit()/removeTransit() which were present before, but now modified and made a bit more robust, more closely matching literature method of transit compartments.

      • addDepot()/removeDepot() which were present before, but modified to be a bit more robust.

      • addWeibullAbs() which adds a Weibull absorption to a PK model

      • convertMM() converts linear elimination to Michaelis-Menten elimination

      • transPK() converts the cl style parameter transformations to various other PK transformations like k, aob, alpha, k12

    • PD model building functions

    • addIndirectLin() -- this adds an indirect effect model to a PK model that has a concentration Cc in the model. This purposely uses a simple linear effect of Cc*Ek or Cc*Ik so it will be easy to parse and turn into other functional forms (like Emax or Hill). If the PK model is not present it will use Cc as a covariate in a purely PD models.

    • addIndirect() -- this builds on addIndirectLin() and adds Emax or Hill models to a PK model. You can also set imax=1 or emax=1 to drop these parameters from being estimated in the model. Additionally hill=TRUE will add a Hill coefficient to the sigmoid model.

    • addEffectCmtLin() -- this adds an effect compartment based on the Cc in the model. The linear effect can be modified into other function forms.

    • addDirectLin() -- this adds a direct effect model based on the Cc in the model.

    • Changing functional forms of Effect models

      • convertEmax() changes linear effect models to Emax models

      • convertEmaxHill() changes linear effect models to Hill models

      • convertQuad() changes linear effect models to quadratic models

      • convertLogLin() changes linear effect models to log-linear models

    • Changing functional forms of Baselines in non-indirect response models

      • addBaselineConst() changes the zero baseline to a estimated constant

      • addBaselineLin() changes the zero baseline to a estimated constant and a linear constant with respect to time.

      • addBaselineExp() changes the zero baseline to a exponential decay with respect to time

      • addBaseline1exp() -- the baseline effect is changed from zero to to an exponential approaching to a constant (with respect to time).

    • Changing model properties (all use addCmtProp())

      • addBioavailability() adds bioavailability property to a compartment

      • addRate() adds a modeled rate to a compartment

      • addDur() adds modeled duration to a compartment

      • addIni() adds an initial value to a compartment

      • addLag() adds a lag time to the a compartment

  • Add Carlsson Petri (2021) liraglutide PK model

  • Add Cirincione (2017) exenatide immediate-release PK model

  • Add a variety of indirect response models

  • Add a variety of tumor growth inhibition models and move all oncology models into a new model database directory

  • Add a variety of double-absorption PK models

  • cp and related cpddSd and cppropSd were renamed to Cc, CcAddSd and CcPropSd (fix #70).

  • Multiple-endpoint models will have the DV column in the modeldb separated by commas.

Version 0.2.0

  • Work with the new rxode2 version 2.0.12 model() and ini() assignment methods.
  • Therapeutic-area specific models have begun being added.
  • Models can now give the user some additional information load via the message meta-data.
  • Models can now be in different directories. The change is for ease of maintaining the library, it is not a change that affects users.
  • A regression where addEta() did not change the parameter, related to a change in rxode2, was fixed.
  • addEta() detects where to add etas more robustly when covariates are on the parameter.

Models added

  • Add Davda (2014) mAb consensus model
  • Add Liu (2017) time-dependent clearance model based on nivolumab
  • Add Kovalenko (2020) dupilumab PK model
  • Add Soehoel (2022) tralokinumab PK model
  • Add Zhu (2017) lebrikizumab PK model

Version 0.1.0

  • Initial version