### Current versions

### News

#### November 24th 2023

**Package PFIM 6.0**

The R package PFIM 6.0 is the result of a complete refactoring of the previous version PFIM 5.0 using the formal object oriented system S4. The software architecture has been completely overhauled to make the code easier to develop and maintain. There are no new features compared to PFIM 5.0, but as a novelty, this new version offers users a completely new and simplest possible usage script to facilitate design evaluation and optimization. The package vignette and the design evaluation and optimisation examples have all been re-coded with this in mind, providing users with a solid foundation of example scripts.

Please use this version PFIM 6.0 instead PFIM 5.0 as some bugs were fixed.

#### June 24th 2022

**Package PFIM 5.0**

The R package PFIM 5.0 is the version of PFIM implemented using the formal object oriented system S4 that defines objects having clear object oriented programming characteristics including class and argument definitions, inheritance, as well as argument checking, instantiation and implementation methods.

Some new features of PFIM 5.0:

- A data summary for design evaluation and optimization is provided. The standard data visualization package ggplot2 is used to display the results in clear graphical form (sensitivity graphs, responses over time and optimal design, SE and RSE obtained with the evaluated or the optimised design)
- The package rmarkdown is used to turn all the results into high quality reports that can be easily shared
- Three new optimization algorithms are added: multiplicative algorithm, PGBO (Population Genetic Based Optimization), PSO (Particle Swarm Optimization), in addition to Simplex and Fedorov-Wynn still available
- Developer documentation on all the methods and classes implemented is provided
- User-friendly vignettes are provided that illustrate the package’s capabilities
- The definition of multiple administrations with different routes (oral, IV bolus, IV infusion) and at different doses is eased
- It is possible to optimize both doses and times (e.g. by using the multiplicative and Fedorov-Wynn algorithms)

Some features available in PFIM 4.0 and not (yet) implemented in PFIM 5.0:

- Models with inter-occasion/within-subject variability (e.g. cross-over designs)
- Models with discrete covariates and power of tests
- Use of prior FIM (for multiple stages design)

#### May 27th 2015

**Version PFIM Interface 4.0**

PFIM Interface 4.0 is the new version of the PFIM graphical user interface. This new Interface version incorporates several new features based on the R script program of PFIM 4.0. For population designs, optimisation can be performed with fixed parameters or fixed sampling times. The Fisher information matrix obtained after evaluation or optimisation can be saved in a file. Additional features for Bayesian designs are now available. The Bayesian Fisher information matrix has been implemented. Design for maximum a posteriori estimation of individual parameters can be evaluated or optimised and the predicted shrinkage is also reported. A new way has been added to specify user-defined models through an R function. It is now possible to visualise the graphs of the model and the sensitivity functions without performing evaluation or optimisation.

#### April 22th 2014

**Version PFIM 4.0**

PFIM 4.0 is an extension of PFIM 3.2.3. This version incorporates several new features. For population designs, optimisation can be performed with fixed parameters or fixed sampling times. The Fisher information matrix obtained after evaluation or optimisation can be saved in a file. Previous information already obtained can be assumed and loaded through a predicted or an observed Fisher information matrix, which is important in the perspective of performing adaptive designs. Additional features for Bayesian designs are now available. The Bayesian Fisher information matrix has been implemented. Design for maximum a posteriori estimation of individual parameters can be evaluated or optimised and the predicted shrinkage is also reported. A new way has been added to specify user-defined models through an R function. It is now possible to visualise the graphs of the model and the sensitivity functions without performing evaluation or optimisation.

#### April 15th 2014

**Version PFIM Interface 3.1.3 (Update, 64bits version + deSolve)**

A new version of PFIM Interface for Windows, PFIM Interface 3.1.3, is now available. This version works both with 32-bit and 64-bit R versions. The R version installed on the user's system can be set through the new option “Path to R”, located in the “Input files” section of the interface.

#### December 19th 2013

**Version PFIM 3.2.3 (Update, 64bits version+deSolve)**

PFIM 3.2.3 works both with 32-bit and 64-bit R versions, and is available in either Windows or Linux operating system.

The “odesolve” package has been replaced by the “deSolve” package, as odesolve is obsolete as R release 2.6.0.

#### March 17th 2011

**Version PFIM Interface 3.1**

PFIM Interface 3.1 is an extension of the version PFIM Interface 2.1 and is based on an extension of the R script version PFIM 3.0 dedicated to design evaluation and optimisation for multiple response models.

This interface version is extended for multiple response models. It can also be used for single response models instead of PFIM interface 2.1 to compute the Fisher information matrix for nonlinear mixed effects models. In addition to this extension, options have been added for model specification and development of the expression of the Fisher information matrix (MF). Regarding model specification, the library of standard PK models has been completed with three compartment models with linear elimination and models with Michaelis-Menten elimination (one, two and three compartment models). Furthermore, a library of pharmacodynamic (PD) models is now available. Concerning the expression of the Fisher information matrix, PFIM Interface 3.1 can handle either a block diagonal Fisher information matrix or a full one.

**Version PFIM 3.2.2** (Release of new library documentation + bug correction for PFIM with inter-occasion variability in case of 4 occasions or more + with discrete covariates on several parameters)

This new update includes a new documentation of the library of models implemented in PFIM 3.2, presenting mathematical expressions for one, two and three compartment pharmacokinetic models with linear elimination or with Michaelis-Menten elimination as well as for pharmacodynamic models.

Furthermore, an error in the previous version concerning the implementation of the block of variance terms for inter-occasion variability (IOV) in the Fisher information matrix, which leads to under-predict the standard errors of IOV in case of 4 occasions or more, has been fixed. A bug concerning discrete covariates when considering the same covariate effect on several parameters has also been corrected. This was an output problem for covariates only and did not impact the criterion.

#### July 12th 2010

**Version PFIM 3.2.1 (change of the file libFED.dll + bug correction in the call of the Fedorov-Wynn algorithm + bug correction in the PD libraries)**

In the previous versions, the length of the character string "directory", which is defined in the file PFIM3.2.r, was limited to 109 characters when using the Federov-Wynn algorithm. Previously, when the length of "directory" was greater than 109 characters, there was the following error message in the R command: "Error in 1:np.f : argument of length 0" or the R software would abnormally terminate. The implementation in the dynamically loaded library of the Federov-Wynn algorithm (*libFED.dll*) has been modified; the length limitation is no longer present in the version of PFIM 3.2.1.

A bug in the call of the Fedorov-Wynn algorithm has also been corrected. The bug concerned the generation of the list of possible elementary designs in the initial step of design optimisation, when the user constrained PFIM to choose one sampling time from a sampling window composed of this sampling time only. The implementation in the file *algofedorov3.2.r* has been modified to take into account this error.

Furthermore, a bug in the previous version of PFIM 3.2 concerning the implementation of the Imax models has been fixed. The bug concerns the three library files *LibraryPD_PDdesign.r*, *LibraryPD_PKPDdesign.r* and *CreateModel_PKPDdesign.r*.

#### January 23th 2010

**Version PFIM 3.2**

PFIM 3.2 is an extension of the version PFIM 3.0. This version incorporates new features in terms of model specification and expression of the Fisher information matrix. Regarding model specification, the library of standard pharmacokinetic (PK) models has been completed by the three compartment models with linear elimination and models with Michaelis-Menten elimination (one, two and three compartment models). Furthermore, a library of pharmacodynamic (PD) models is now available. Concerning the expression of the Fisher information matrix, PFIM 3.2 can handle either a block diagonal Fisher information matrix or the complete one. It is now also possible in PFIM 3.2 to use models including inter-occasion variability (IOV) with replicated designs at each occasion. Last, a new feature of PFIM 3.2 is the computation of the Fisher information matrix for models including fixed effects for the influence of discrete covariates on the parameters. It can be specified if covariates change or not through the different occasions. The computation of the predicted power of the Wald test for comparison or equivalence test for a given distribution of a discrete covariate as well as the number of subjects needed to achieve a given power can be computed.

#### May 20th 2008

**Update of version PFIM Interface 2.1 (bug correction)**

A bug in the previous version of the Fedorov-Wynn algorithm has been corrected.

The bug concerns the computation of the number of subjects in the optimised designs using the Fedorov-Wynn algorithm (designs optimised with the Simplex algorithm are unaffected by the bug).

The only results affected are population designs combining elementary designs with unequal number of sampling times. eg :

- if the final design is for instance ((1,2,3),(1,3,5)) with frequencies (alpha1, alpha2) it is unaffected by the bug
- if the final design is for instance ((1,2,3),(1,3)) with frequencies (alpha1, alpha2), it is affected by the bug, and the frequencies will not be correct. In this case the correct frequencies should be respectively (alpha1/3)/sum and (alpha2/2)/sum where sum=(alpha1/3)+(alpha2/2) to normalise the sum of the frequencies to 1. ("frequencies" refer to the frequencies reported by the bugged version of the algorithm)

Because of this bug, the final criterion and estimates of SE for the parameters in the model, being computed with the wrong frequencies, is also incorrect, but differences should be small and unlikely to be clinically significant. The new version of the Fedorov-Wynn outputs the correct frequencies and we encourage users to update PFIM interface 2.1 with the new version.

#### April 24th 2008

**Version PFIM 3.0**

PFIM 3.0 is an extension of PFIM for multiple response models. It is based on extensions of R functions PFIM1.2 and PFIMOPT1.0 for evaluation and optimisation respectively. Conversely to the previous version, only one function is proposed allowing both evaluation and optimisation of population designs. Options have been added for model specification and optimisation, compared to PFIM 1.2 and PFIMOPT 1.0. The model can be written using an analytical form or using a differential equation system. Moreover, with these new versions a library of pharmacokinetics models is available. Regarding optimisation step, the Federov-Wynn algorithm has been added as an alternative to the Simplex algorithm. It allows to optimise design with fixed sampling times in opposite to the Simplex. Moreover, it considers only pre-specified sampling times, avoiding, clinically unfeasible sampling times.

#### January 29th 2008

**Version PFIM Interface 2.1 (corrected version of PFIM Interface 2.0)**

PFIM Interface 2.1 allows both evaluation and optimisation and it is was programmed as a graphical user interface package using the R software. It is based on the previous functions PFIM1.2 and PFIMOPT1.0. Documentation on this graphical user interface version is available in the relevant section and includes detailed explanations and examples as to how to use **PFIM Interface 2.1**. Options have been added for model specification and optimisation, compared to PFIM 1.2 and PFIMOPT 1.0. The model can be written using an analytical form or using a differential equation system. Moreover, with these new versions a library of pharmacokinetics models is available. Regarding optimisation step, the Federov-Wynn algorithm has been added as an alternative to the Simplex algorithm. It allows to optimise design with fixed sampling times in opposite to the Simplex. Moreover, it considers only pre-specified sampling times, avoiding, clinically unfeasible sampling times.

#### June 2007

**Version PFIM Interface 2.0 (working version)**

#### 2003

**Version PFIM 1.2**

PFIM 1.2 is a R function for evaluation of population designs.

**Version PFIMOPT 1.0**

PFIMOPT 1.0 is a R function for optimisation of population designs.

#### 2001

**Version PFIM 1.1**