WNL Classic Models

WinNonlin Classic models include the following. Many of the WinNonlin Models can be run using the NLME engine (even if you do not have an NLME license). This is done by setting up a Phoenix Model object for individual modeling and using the Set WNL Model button to select the model. See “PK model options” in the Phoenix NLME documentation for more information. Refer to “An example of individual modeling with Phoenix model” for an illustration of how a dataset can be fitted to a two-com­partment model with first-order absorption in the pharmacokinetic model library using either the Win­Nonlin PK Model or a Phoenix Model GUI.

Note:There can be a loss of accuracy in WNL Classic Modeling univariate confidence intervals for small sample sizes (NDF < 5). The Univariate CIs in WNL use an approximation for the t-value which is very accurate when the degrees of freedom is at least five, but loses accuracy as the degrees of freedom approaches one. The degrees of freedom are the number of observations minus the number of parameters being estimated (not counting parameters that are within the singularity tol­erance, i.e., nearly completely correlated).

In extremely rare instances, the nonlinear modeling core computational engine may get into an infinite loop during the minimization process. This infinite looping will cause Phoenix to “hang” and the application must be shutdown using the Task Manager. The process wnlpk32.exe may also need to be shutdown. The problem typically occurs when the parameter space in which the program is working within is very flat. To work around the problem, it is first suggested that the user change the minimization algorithm found on the Engine Settings tab to Nelder-Mead and retry the problem. If this fails to correct the problem, varying the initial estimates and/or using bounds on the parameters may allow processing to complete as expected.

The Classic Model interface

Use one of the following to add a WinNonlin Classic Model object to a Workflow:

Right-click menu for a Workflow object: New > NCA and Toolbox > WNL6 Classic Modeling > [object name].
Main menu: Insert > NCA and Toolbox > WNL6 Classic Modeling > [object name].
Right-click menu for a worksheet: Send To > NCA and Toolbox > WNL6 Classic Modeling > [object name].

Note:To view the object in its own window, select it in the Object Browser and double-click it or press ENTER. All instructions for setting up and execution are the same whether the object is viewed in its own window or in Phoenix view.

Main Mappings panel

Use the Main Mappings panel to identify how input variables are used in a model. A separate analysis is performed for each profile. Required input is highlighted orange in the interface.

None: Data types mapped to this context are not included in any analysis or output.

Sort: Categorical variable(s) identifying individual data profiles, such as subject ID and treat­ment in a crossover study. A separate analysis is done for each unique combination of sort variable values.

X: For Dissolution, Indirect Response, Linear, PD, and PK/PD Linked Models. The indepen­dent variable in a dataset.

Y: For Dissolution, Indirect Response, Linear, PD, and PK/PD Linked Models. The dependent variable a dataset. Not needed if the Simulation checkbox is selected.

Time: For Michaelis-Menten, PK, and ASCII Models. Nominal or actual time collection points in a study.

Concentration: For Michaelis-Menten, PK Models, and ASCII Models. Drug concentration values in the blood. Not needed if the Simulation checkbox is selected.

Function: For ASCII Models. Model function.

Carry: Data variable(s) to include in the output worksheets. Note that time-dependent data variables (those that change over the course of a profile) are not carried over to time-indepen­dent output (e.g., Final Parameters), only to time-dependent output (e.g., Summary).

Weight: For Indirect Response, PK, PD, and PK/PD Linked Models. Used if weighting data is contained in the dataset. This column is only available if User Defined and Source are selected in the Weighting/Dosing Options tab.

Note:When using a PK operational object, external worksheets for stripping dose, units and initial esti­mates can be accessed in different ways. The differences will occur if there is more than 1 row of information on these external worksheets that correspond to one or more individual profiles of data of the Main input worksheet.

In such cases, the stripping dose for PK models will be determined as the first value found on that external worksheet whereas the units and initial parameters will be based on the last row found on those external worksheets (for any given profile). To avoid any confusion stemming from these dif­ferences, it is suggested that external worksheets maintain a one-to-one row-based correspon­dence to the Main input profiles whenever possible.

Dosing panel

Available for Indirect Response, PK, and PK/PD Linked Models only, the Dosing panel allows users to type or map dosing data for the different models. The Dosing panel mapping columns change depending on the PK model type selected in the Model Selection tab. Required input is highlighted orange in the interface.

If multiple sort variables have been mapped in the Main Mappings panel, the Select sorts dialog is dis­played so that the user can select the sort variables to include in the internal worksheet. Required input is highlighted orange in the interface.

None: Data types mapped to this context are not included in any analysis or output.

Sort: Sort variables.

Time: The time of dose administration.

Dose: The amount of drug administered. Dosing units are used.

End Time: The end time of the infusion.

Infusion Length: The total amount of time for an IV infusion. Only used in conjunction with a bolus dose.

Bolus: Amount of the bolus dose.

Amount Infused: Total amount of drug infused. Dosing units are used.

Note:If the dosing data is entered using the internal dosing worksheet, and different profiles require dif­ferent numbers of doses, then leave the Time, End Time, Dose, Infusion Length, Bolus, or Amount Infused cells blank for profiles that require less than the highest number of doses.

When using an internal worksheet, click the Rebuild button to reset the worksheet to its default state and delete all entered values.

Initial Estimates panel

The Initial Estimates panel allows users to type or map initial values and lower and upper boundaries for different indirect response parameters.

Note:Multiple-dose datasets require users to provide initial parameter values. For more on setting initial parameter estimates, see the “Parameter Estimates and Boundaries Rules” section.

If initial estimates and parameter boundaries for ASCII models are not set in the code, then they must be set in the Initial Estimates panel.

In datasets containing multiple sort variables, the initial estimates must be provided for each level of the sort variables unless the Propagate Final Estimates checkbox is selected. Checking this box applies the same types of parameter calculations and boundaries to all sort levels. Required input is highlighted orange in the interface.

If multiple sort variables have been mapped in the Main Mappings panel, the Select sorts dialog is displayed so that the user can select the sort variables to include in the internal worksheet.

None: Data types mapped to this context are not included in any analysis or output.

Sort: Sort variables.

Parameter: Model parameters such as v (volume) or km (Michaelis constant).

Initial: Initial parameter estimate values.

Fixed or Estimated: For Dissolution models. Whether the initial parameter is fixed or esti­mated.

Lower: Lower parameter boundary.

Upper: Upper parameter boundary.

The lower and upper values limit the range of values the parameters can take on during model fit­ting. This can be very valuable if the parameter values become unrealistic or the model will not converge. Although bounds are not always required, it is recommended that Lower and Upper bounds be used routinely. If they are used, every parameter must be given lower and upper bounds.

The Phoenix default bounds are zero for one bound and ten times the initial estimate for the other bound. For models with parameters that may be either positive or negative, user-defined bounds are preferred.

Note:For the WinNonlin Generated Initial Parameter Values option with Dissolution models, to avoid getting pop-up warnings that “WinNonlin will determine initial estimate” when using the Initial Esti­mates internal worksheet setup, delete the initial values and change the menu option from Esti­mated to Fixed before entering the initial estimates. Once the dropdown is changed from Fixed to Estimated, the initial value entered cannot be deleted and the warning pop-up will be displayed. Note that this situation does not affect the estimation process, as the entered initial value will not be used and WinNonlin will estimate the initial value as requested.

When using an internal worksheet, click the Rebuild button to reset the worksheet to its default state and delete all entered values.

PK Parameters panel

This panel is available for Indirect Response and PK/PD Linked models only.

Note:Users are required to enter initial PK parameter values in the PK Parameters panel in order for the model to run.

To display the units in the PK Parameters panel, units must be included in the time, concentration, and dose input data and the concentration units entered in the PK Units text field in the Model Selection tab.

If multiple sort variables have been mapped in the Main Mappings panel, the Select sorts dialog is dis­played so that the user can select the sort variables to include in the internal worksheet. Required input is highlighted orange in the interface.

None: Data types mapped to this context are not included in any analysis or output.

Sort: Sort variables.

Parameter: Name of the parameter.

Value: Value of the parameter.

When using an internal worksheet, click the Rebuild button to reset the worksheet to its default state and delete all entered values.

Units panel

For all WNL Classic Models except Michaelis-Menten, an object’s display units can be changed to fit a user’s preferences. Required input is highlighted orange in the interface.

Depending on the type of model, there are some prerequisites for setting preferred units:

For Indirect Response and PK Models, the Time, Concentration, and Dose data must all contain units before users can set preferred units.

For PD and PK/PD Linked Models, the data mapped to the X and Y contexts must all contain units before users can set preferred units.

For ASCII Models, units must be set in the ASCII model code.

Each parameter used in a model and the parameter’s default units are listed in the Units panel.

None: Data types mapped to this context are not included in any analysis or output.

Name: Model parameters associated with the units.

Default: The model object’s default units.

Preferred: The user’s preferred units for the parameter.

When using an internal worksheet, click the Rebuild button to reset the worksheet to its default state and delete all entered values.

Note:if you see an “Insufficient units” message in the table, check that units are defined for time and concentration in your input.

Stripping Dose panel

The Stripping Dose panel is available when one of the macro constant PK models that use a stripping dose is specified in the Model Selection tab (i.e., models 8, 13, 14, 17, or 18). This panel is used to enter the stripping dose amount, which is the dose associated with initial parameter values for macro constant models.

If a user selects a macro constant PK model and provides user-specified initial estimates, then the user must specify the associated dose. If a user chooses to have Phoenix generate initial parameter values, then the stripping dose is identical to the administered dose.

None: Data types mapped to this context are not included in any analysis or output.

Sort: Sort variables.

Dose: The stripping dose.

If multiple sort values have been mapped in the Main Mapping panel, the Select sorts dialog is dis­played so that the user can select the sort variables to include in the internal worksheet.

Constants panel

The Constants panel is available only for Michaelis-Menten and ASCII Models and allows users to type or map dosing data for the model. Specific dosing information is determined using dosing con­stants. For more on how constants relate to dosing in Michaelis-Menten models, see “Dosing con­stants for the Michaelis-Menten model”. For more on how constants relate to dosing in ASCII models, see “ASCII Model dosing constants”.

When using an external worksheet for Constants, the Number of Constants will be determined from the number of rows per profile after mapping the following:

None: Data types mapped to this context are not included in any analysis or output.

Sort: Sort variables.

Order: The number of dosing constants used.

Value: The value for each dosing constant. For example, the value for CON[0] is 1 if the model is single-dose.

For an internal worksheet, set the Number of Constants on the Options panel, or specify NCON in the model text, to expand the internal worksheet for entering the Constants.

Format panel

The Format panel is only available for ASCII Models and is used to map ASCII code to an ASCII model. Users can view and edit ASCII model code in this panel.

Click Yes to have the ASCII code copied to the internal text editor.

Click No to remove the ASCII code and start with a blank internal text editor.

If ASCII code was previously mapped to the WNL5 ASCII Format panel then that code is displayed in the internal text editor, instead of a blank panel.

More information on the text editor is available on the Syncfusion website.

Model Selection tab

The Model Selection tab for most of the WNL Classic Models allows users to select a model and whether or not the model uses simulated data/clearance parameter. (For the ASCII Model, the Model Selection tab contains weighting options as described in the “Weighting/Dosing Options tab” descrip­tion.)

IRmodelSelectiontab.png 

Refer to any of the following sections for model details:

Linear models

Michaelis-Menten models

Pharmacodynamic models

Pharmacokinetic models

Selecting a model displays a diagram beside the model selection menu that describes the model’s functions. The model’s equation is listed beneath the diagram.

PKmodel1diagandopts.png 

For Indirect Response, PK, and PK/PD Linked models, select the Clearance checkbox to add a clearance parameter to the model. (The clearance parameter option is not available for PK models 8, 10, 13, 14, 17, 18, or 19.)

For Michaelis-Menten Models, use the Number of Constants box to type or select the num­ber of dosing constants used per profile. (For more on how the number of constants corre­sponds to the number, amount, and time of doses, see “Dosing constants for the Michaelis-Menten model”.)

For PK/PD Linked Models, enter concentration units in the PK Units text field or click the Units Builder UnitsBuilderbutton_1.png button to use the Units Builder dialog.

Note:For PK/PD Linked Models, to view all PK parameter units in the PK Parameters panel, users must supply concentration units in the PK Units text field in the Model Selection tab and dose units in the Weighting/Dosing Options tab.

PKmodel1optnsselected.png 

Linked Model tab

The Linked Model tab allows users to select the model to link with model specified in the Model Selec­tion tab. It is available only for the Indirect Response and PK/PD Linked Models.

IRmodelLinkedModeltab.png 

IRmodel51diagram.png 

Weighting/Dosing Options tab

The Weighting Options tab (in Dissolution, Linear, M-M, and PD Model objects) and the Weighting/Dosing Options tab (in Indirect Response, PK, and PK/PD Linked Model objects) allows users to select a weighting scheme and specify and preview dosing options.

IRmodelWghtDosingOptstab.png 

Weighting options

User Defined: Weights are read from a column in the dataset.

Uniform: Users can enter custom observed to or predicted to the power of N values. If selected, then users must select Observed or Predicted in the Source menu and type the power value in the Power to text field.

1/Y: Weight the data by 1/observed Y.

1/Yhat: Weight the data by 1/predicted Y (iterative reweighting).

1/(Y*Y): Weight the data by 1/observed Y2.

1/(Yhat*Yhat): Weight the data by 1/predicted Y2 (iterative reweighting).

Source: Selecting this option sets the weighting to User Defined and adds a Weight column to the Main Mappings panel. If selected, users must map the weighting column in the dataset to the Weight context in the Main Mappings panel.

Observed: Select to use weighted least squares. This is the default selection for 1/Y and 1/(Y*Y). The default power for 1/Y is –1 and for 1/(Y*Y) it is –2.

Predicted: Select to use iterative reweighting. This is the default selection for 1/Yhat and 1/(Yhat*Yhat). The default power for 1/Yhat is –1 and for 1/(Yhat*Yhat) it is –2.

Entering -1 automatically sets the weighting to 1/Y (if Observed is the source) or 1/Yhat (if Predicted is the source).

Entering -2 automatically sets the weighting to 1/(Y*Y) (if Observed is the source) or 1/(Yhat*Yhat) (if Predicted is the source).

Dosing options

Available for Indirect Response, PK, and PK/PD Linked models.

Note:For Indirect Response Models, to view all PK parameter units in the PK Parameters panel, users must supply units for the time and concentration data in the input and specifying dose units in the Weighting/Dosing Options tab.

For PK/PD Linked Models, to view all PK parameter units in the PK Parameters panel, users must supply concentration units in the PK Units text field in the Model Selection tab and dose units in the Weighting/Dosing Options tab.

If doses are in milligrams per kilogram of body weight, select mg as the dosing unit and kg as the dose normalization.

The Normalization menu affects the output parameter units. For example, if dose volume is in liters, selecting kg as the dose normalization changes the units to L/kg.

Dose normalization affects units for all volume and clearance parameters in PK models.

Parameter Options tab

All iterative estimation procedures require initial estimates of the parameter values. Phoenix com­putes initial estimates via curve stripping for single-dose models. For all other situations, including multiple-dose models, users must provide initial estimates or boundaries to be used by Phoenix in creating initial estimates. Parameter boundaries provide a basis for grid searching initial parameter estimates, and also limit the estimates during modeling. This is useful if the values become unrealistic or the model does not converge. For more on setting initial parameter estimates, refer to “Parameter Estimates and Boundaries Rules”.

IRmodelParameterOptstab.png 

Set the parameter calculation method:

Note:The default minimization method, Gauss-Newton (Hartley) (located in the Engine Settings tab), and the Parameter Boundaries option Do Not Use Bounds are recommended for all Linear mod­els.

If the User Supplied Bounds option button is selected, Phoenix uses curve stripping to pro­vide initial estimates. If curve stripping fails, then Phoenix uses the grid search method.

If the WinNonlin Bounds option button is selected, Phoenix uses curve stripping to provide initial estimates, and then applies boundaries to the model parameters for model fitting. If curve stripping fails, the model fails because Phoenix cannot use grid search for initial esti­mates without user-supplied boundaries.

Set the boundary calculation method:

Parameter boundaries provide a basis for grid searching initial parameter estimates, and also limit the estimates during modeling. This is useful if the values become unrealistic or the model does not con­verge. For more on using parameter boundaries, refer to “Parameter Estimates and Boundaries Rules”.

Engine Settings tab

The Engine Settings tab provides control over the model fitting algorithm and related settings.

IRmodelEngineSettingstab.png 

Note:The use of bounds is recommended with Methods 2 and 3. For linear regressions, use Method 3 without bounds.

pkmodels02416.png 

(1)

Note:To better reflect peaks (for IV dosing) and troughs (for extravascular, IV infusion and IV bolus dos­ing), the predicted data for the built-in PK models includes dosing times, in addition to the concen­trations generated. For all three types, concentrations are generated at dosing times; in addition, for infusion models, data are generated for the end of infusion.

500 for Nelder-Mead minimization. Each iteration is a reflection of the simplex.

50 for Gauss-Newton (Levenberg and Hartley) or Gauss-Newton (Hartley) minimizations.

Plots tab

IRmodelPlotstab.png 

Results

Note:This section is meant to provide guidance and references to aid in the interpretation of modeling output, and is not a replacement for a PK or statistics textbook.

After a Classic model is run, the output is displayed on the Results tab in Phoenix. The output is dis­cussed in the following sections.

Core Output: text version of all model settings and output, including any errors that occurred during modeling. See “Core Output File” for a full description.

Settings: test version of all user-defined settings.

Worksheet output: worksheets listing input data, modeling iterations and output parameters, as well as several measures of fit.

Plot output: plots of observed and predicted data, residuals, and other quantities, depending on the model run.

Worksheet output

Worksheet output contains summary tables of the modeling data and a summary of the information in the Core Output. The worksheets generated depend on the analysis type and model settings. They present the output in a form that can be used for reporting and further analyses and are listed on the Results tab underneath Output Data.

PK models output worksheets

Name

Description

Condition Numbers

Rank and condition number of the matrix of partial derivatives for each iteration.
The matrix is of full rank, since Rank is equal to the number of parameters.
If the Rank were less than three, that would indicate that there was not enough information in the data to estimate all three parameters.
The condition value is the square root of the ratio of the largest to the smallest eigenvalue and values should be less than 10^n where n is the number of parameters.

Correlation Matrix

A correlation matrix for the parameters, for each sort level.
If any values get close to 1 or –1, there may be too many parame­ters in the model and a simpler model may work better.

Diagnostics

Diagnostics for each function in the model and for the total:
- corrected sum of squared observations (CSS)
- weighted corrected sum of squared observations (WCSS)
- sum of squared residuals (SSR)
- weighted sum of squared residuals (WSSR)
- estimate of residual standard deviation (S)
- degrees of freedom (DF)
- correlation (CORR_(OBS,PRED)) between observed Y and pre­dicted Y
- weighted correlation (WT_CORR_(OBS,PRED))
- Akaike Information Criterion (AIC) goodness of fit measurement
- Schwarz Bayesian Criterion (SBC) goodness of fit measure­menta

Differential Equations

The value of the partial derivatives for each parameter at each time point for each value of the sort variables.

Dosing Used

The dosing regimen specified for the modeling.

Eigenvalues

Eigenvalues for each level of the sort variables.
(An eigenvalue of matrix A is a number l, such that Ax=lx for some vector x, where x is the eigenvector. Eigenvalues and their associated eigenvectors can be thought of as building blocks for matrices.)

Final Parameters and Final Parameters Pivoted

Parameter names, units, estimates, standard error of the esti­mates, CV% (values < 20% are generally considered to be very good), univariate intervals, and planar intervals for each level of the sort variables.

Fitted Values

(Dissolution models) Predicted data for each profile.

Initial Estimates

Parameter names, initial values, and lower and upper bounds for each level of the sort variables.

Minimization Process

Iteration number, weighted sum of squares, and value for each parameter, for each level of the sort variables.
This worksheet shows how parameter values converged as the iterations were performed.
If the number of iterations is approaching the specified limit, there may be some problems with the model.

Parameters

(Dissolution models) The smoothing parameter delta and absorp­tion lag time for each profile.

Partial Derivatives and Stacked Partial Derivatives

Values of the differential equations at each time in the dataset.

Predicted Data

Time and predicted Y for multiple time points, for each sort level.

Secondary Parameters and Secondary Parameters Pivoted

Available for Michaelis-Menten, PK, PD, PK/PD Linked and ASCII models.
Secondary parameter name, units, estimate, standard error of the estimate, and CV% for each sort level.

Summary Tableb 

The sort variables, X, Y, transformed X, transformed Y, predicted Y, residual, weight, standard error of predicted Y, standardized residuals, for each sort level.
For link models, also includes CP and Ce.
For indirect response models, also includes CP.

Values

(Dissolution models) Time, input rate, cumulative amount (Cumul_Amt, using the dose units) and fraction input (Cumul_Amt/test dose or, if no test doses are given, then fraction input approaches one) for each profile.

Variance Covariance Matrix

A variance-covariance matrix for the parameters, for each sort level.

User Settings

Model number, minimization method, convergence criterion, maxi­mum number of iterations allowed, and the weighting scheme.

aAIC and SBC are only meaningful during comparison of models. A smaller value is better, negative is better than positive, and a more negative value is even better. AIC is computed as:
AIC=N log (WRSS)+2P, where N is the number of observations with positive weight, log is the nat­ural logarithm, WRSS is the weighted residual sum of squares, P is the number of parameters.

b If there are no statements to transform the data, then X and Y will equal X(obs) and Y(obs).

.

Plot output

Analysis produces up to eight graphs that are divided by each level of the sort variable. Plot output is listed underneath Plots in the Results tab.

Users can double-click any plot in the Results tab to edit it. (See the “Plot Options tab” description for editing options.)


Last modified date:9/10/19
Certara USA, Inc.
Legal Notice | Contact Certara
© 2019 Certara USA, Inc. All rights reserved.