Define relationship between parameters and covariates
CovariateModel defines the relationship between
estimated parameters and covariates.
Tip
Use a CovariateModel object as an input argument
to sbiofitmixed to fit a model
with covariate dependencies. Before using the CovariateModel object,
set the FixedEffectValues property
to specify the initial estimates for the fixed effects.
CovModelObj = CovariateModelCovariateModel object.
CovModelObj = CovariateModel(Expression)CovariateModel object with its
Expression property set to Expression,
a character vector, string, string vector, or cell array of character vectors, where
each character vector or string represents the relationship between a parameter being
estimated and one or more covariates. Expression must denote
fixed effects with the prefix theta, and random effects with the
prefix eta. Each character vector or string in
Expression must be in the form:
parameterName = relationship |
This example of an expression defines the relationship between
a parameter (volume) and a covariate (weight),
with fixed effects, but no random effects:
|
If a model component name or covariate name is not a valid
MATLAB® variable name, surround it by square brackets when referring to it in the
expression. For example, if the name of a species is DNA
polymerase+, write [DNA polymerase+]. If a covariate name
itself contains square brackets, you cannot use it in the expression.
This table illustrates expression formats for some common parameter-covariate relationships.
| Parameter-Covariate Relationship | Expression Format |
|---|---|
| Linear with random effect | Cl = theta1 + theta2*WEIGHT + eta1 |
| Exponential without random effect | Cl = exp(theta_Cl + theta_Cl_WT*WEIGHT) |
| Exponential, WEIGHT centered by mean, has random effect | Cl = exp(theta1 + theta2*(WEIGHT - mean(WEIGHT)) +
eta1) |
| Exponential, log(WEIGHT), which is equivalent to power model | Cl = exp(theta1 + theta2*log(WEIGHT) + eta1) |
| Exponential, dependent on WEIGHT and AGE, has random effect | Cl = exp(theta1 + theta2*WEIGHT + theta3*AGE + eta1) |
| Inverse of probit, dependent on WEIGHT and AGE, has random effect | Cl = probitinv(theta1 + theta2*WEIGHT + theta3*AGE
+ eta1) |
| Inverse of logit, dependent on WEIGHT and AGE, has random effect | Cl = logitinv(theta1 + theta2*WEIGHT + theta3*AGE
+ eta1) |
Tip
To simultaneously fit data from multiple dose levels, use a CovariateModel object
as an input argument to sbiofitmixed,
and omit the random effect (eta) from the Expression property
in the CovariateModel object.
| constructDefaultFixedEffectValues (covmodel) | Create initial estimate vector needed for fit |
| verify (covmodel) | Check covariate model for errors |
| CovariateLabels (CovariateModel) | Labels for covariates in CovariateModel object |
| Expression (CovariateModel) | Define relationship between parameters and covariates |
| FixedEffectDescription (CovariateModel) | Descriptions of fixed effects in CovariateModel object |
| FixedEffectNames (CovariateModel) | Names of fixed effects in CovariateModel object |
| FixedEffectValues (CovariateModel) | Values for initial estimates of fixed effects in CovariateModel object |
| ParameterNames (CovariateModel) | Names of parameters in CovariateModel object |
| RandomEffectNames (CovariateModel) | Names of random effects in CovariateModel object |
Create a CovariateModel object and set the Expression property
to define the relationship between two parameters (clearance and volume)
and two covariates (weight and age) using fixed effects (thetas)
and random effects (etas):
covModelObj = CovariateModel
covModelObj.Expression = {'CL = theta1 + theta2*WT + eta1', 'V = theta3 + theta4*AGE + eta2'};construct | getCovariateData | PKData object | PKModelDesign
object | sbionlmefit | sbionlmefitsa