The examples below require the package campismod.
library(campsismod)The examples below are illustrated based on the reference 2-compartment PK model that you find in the built-in library. This model has 5 parameters. All these parameters have inter-individual variability defined.
model <- model_suite$pk$`2cpt_fo`
model## [MAIN]
## TVKA=THETA_KA
## TVVC=THETA_VC
## TVVP=THETA_VP
## TVQ=THETA_Q
## TVCL=THETA_CL
## TVPROP_RUV=THETA_PROP_RUV
##
## KA=TVKA * exp(ETA_KA)
## VC=TVVC * exp(ETA_VC)
## VP=TVVP * exp(ETA_VP)
## Q=TVQ * exp(ETA_Q)
## CL=TVCL * exp(ETA_CL)
## PROP_RUV=TVPROP_RUV
##
## [ODE]
## d/dt(A_ABS)=-KA*A_ABS
## d/dt(A_CENTRAL)=KA*A_ABS + Q/VP*A_PERIPHERAL - Q/VC*A_CENTRAL - CL/VC*A_CENTRAL
## d/dt(A_PERIPHERAL)=Q/VC*A_CENTRAL - Q/VP*A_PERIPHERAL
##
## [ERROR]
## CONC=A_CENTRAL/VC
## if (CONC <= 0.001) CONC=0.001
## IPRED=log(CONC)
## W=PROP_RUV
## Y=IPRED + W*EPS_RUV_FIX
##
##
## THETA's:
## name index value fix
## 1 KA 1 1.0 FALSE
## 2 VC 2 60.0 FALSE
## 3 VP 3 10.0 FALSE
## 4 Q 4 2.0 FALSE
## 5 CL 5 3.0 FALSE
## 6 PROP_RUV 6 0.1 FALSE
## OMEGA's:
## name index index2 value fix type same
## 1 KA 1 1 25 FALSE cv% NA
## 2 VC 2 2 25 FALSE cv% NA
## 3 VP 3 3 25 FALSE cv% NA
## 4 Q 4 4 25 FALSE cv% NA
## 5 CL 5 5 25 FALSE cv% NA
## SIGMA's:
## name index index2 value fix type
## 1 RUV_FIX 1 1 1 TRUE var
## No variance-covariance matrix
##
## Compartments:
## A_ABS (CMT=1)
## A_CENTRAL (CMT=2)
## A_PERIPHERAL (CMT=3)To retrieve a parameter by its name, just call the method
find as follows:
model %>% find(Theta("CL"))## name index value fix
## 1 CL 5 3 FALSEmodel %>% find(Omega("KA"))## name index index2 value fix type same
## 1 KA 1 1 25 FALSE cv% NAmodel %>% find(Sigma("RUV_FIX"))## name index index2 value fix type
## 1 RUV_FIX 1 1 1 TRUE varThese parameters can also alternatively be retrieved by their
index(es). Use the specific method getByIndex created for
that purpose:
model@parameters %>% getByIndex(Theta(index=5))## name index value fix
## 1 CL 5 3 FALSEmodel@parameters %>% getByIndex(Omega(index=1, index2=1))## name index index2 value fix type same
## 1 KA 1 1 25 FALSE cv% NAmodel@parameters %>% getByIndex(Sigma(index=1, index2=1))## name index index2 value fix type
## 1 RUV_FIX 1 1 1 TRUE varAccessing parameter values is straightforward. Parameters have a slot
value that may be accessed.
thetaCL <- model %>% find(Theta("CL"))
thetaCL@value## [1] 3For OMEGA and SIGMA parameters, be careful; the interpretation of
this value depends on the type of the parameter. It may be
var (for a variance), covar (for a
covariance), sd (for standard deviation), cv
(value expressed as coefficient of variation), cv% (value
expressed as coefficient of variation in percentage).
For a quick access to the value as variance or covariance, the method
standardise can be called first on the parameter itself.
This is especially useful for values expressed in CV or in standard
deviation.
theta <- Omega(name="TEST", index=1, index2=1, value=15, type="cv%")
theta@value # 15 is returned## [1] 15theta_standardised <- theta %>% standardise() # Conversion to variance
theta_standardised ## name index index2 value fix type same
## 1 TEST 1 1 0.02225061 FALSE var NAtheta_standardised@value## [1] 0.02225061Parameters can be replaced easily. Here are a few examples:
model <- model %>% replace(Theta("KA", value=2)) # Previous value for KA was 1
model <- model %>% replace(Omega("CL", value=20, type="cv%")) # Previous value was a 25% CV
model## [MAIN]
## TVKA=THETA_KA
## TVVC=THETA_VC
## TVVP=THETA_VP
## TVQ=THETA_Q
## TVCL=THETA_CL
## TVPROP_RUV=THETA_PROP_RUV
##
## KA=TVKA * exp(ETA_KA)
## VC=TVVC * exp(ETA_VC)
## VP=TVVP * exp(ETA_VP)
## Q=TVQ * exp(ETA_Q)
## CL=TVCL * exp(ETA_CL)
## PROP_RUV=TVPROP_RUV
##
## [ODE]
## d/dt(A_ABS)=-KA*A_ABS
## d/dt(A_CENTRAL)=KA*A_ABS + Q/VP*A_PERIPHERAL - Q/VC*A_CENTRAL - CL/VC*A_CENTRAL
## d/dt(A_PERIPHERAL)=Q/VC*A_CENTRAL - Q/VP*A_PERIPHERAL
##
## [ERROR]
## CONC=A_CENTRAL/VC
## if (CONC <= 0.001) CONC=0.001
## IPRED=log(CONC)
## W=PROP_RUV
## Y=IPRED + W*EPS_RUV_FIX
##
##
## THETA's:
## name index value fix
## 1 KA 1 2.0 FALSE
## 2 VC 2 60.0 FALSE
## 3 VP 3 10.0 FALSE
## 4 Q 4 2.0 FALSE
## 5 CL 5 3.0 FALSE
## 6 PROP_RUV 6 0.1 FALSE
## OMEGA's:
## name index index2 value fix type same
## 1 KA 1 1 25 FALSE cv% NA
## 2 VC 2 2 25 FALSE cv% NA
## 3 VP 3 3 25 FALSE cv% NA
## 4 Q 4 4 25 FALSE cv% NA
## 5 CL 5 5 20 FALSE cv% NA
## SIGMA's:
## name index index2 value fix type
## 1 RUV_FIX 1 1 1 TRUE var
## No variance-covariance matrix
##
## Compartments:
## A_ABS (CMT=1)
## A_CENTRAL (CMT=2)
## A_PERIPHERAL (CMT=3)Parameters can be deleted. Please note that it does not do anything to the equations. Also, the indexes won’t be re-adjusted. Here are a few examples:
model <- model %>% delete(Theta("KA"))
model <- model %>% delete(Omega("CL"))
model## [MAIN]
## TVKA=THETA_KA
## TVVC=THETA_VC
## TVVP=THETA_VP
## TVQ=THETA_Q
## TVCL=THETA_CL
## TVPROP_RUV=THETA_PROP_RUV
##
## KA=TVKA * exp(ETA_KA)
## VC=TVVC * exp(ETA_VC)
## VP=TVVP * exp(ETA_VP)
## Q=TVQ * exp(ETA_Q)
## CL=TVCL * exp(ETA_CL)
## PROP_RUV=TVPROP_RUV
##
## [ODE]
## d/dt(A_ABS)=-KA*A_ABS
## d/dt(A_CENTRAL)=KA*A_ABS + Q/VP*A_PERIPHERAL - Q/VC*A_CENTRAL - CL/VC*A_CENTRAL
## d/dt(A_PERIPHERAL)=Q/VC*A_CENTRAL - Q/VP*A_PERIPHERAL
##
## [ERROR]
## CONC=A_CENTRAL/VC
## if (CONC <= 0.001) CONC=0.001
## IPRED=log(CONC)
## W=PROP_RUV
## Y=IPRED + W*EPS_RUV_FIX
##
##
## THETA's:
## name index value fix
## 1 VC 2 60.0 FALSE
## 2 VP 3 10.0 FALSE
## 3 Q 4 2.0 FALSE
## 4 CL 5 3.0 FALSE
## 5 PROP_RUV 6 0.1 FALSE
## OMEGA's:
## name index index2 value fix type same
## 1 KA 1 1 25 FALSE cv% NA
## 2 VC 2 2 25 FALSE cv% NA
## 3 VP 3 3 25 FALSE cv% NA
## 4 Q 4 4 25 FALSE cv% NA
## SIGMA's:
## name index index2 value fix type
## 1 RUV_FIX 1 1 1 TRUE var
## No variance-covariance matrix
##
## Compartments:
## A_ABS (CMT=1)
## A_CENTRAL (CMT=2)
## A_PERIPHERAL (CMT=3)As expected, this model will not be valid anymore:
tryCatch({validObject(model, complete=TRUE)}, error=function(msg) {
print(msg)
})## <simpleError in validObject(model, complete = TRUE): objet de classe "campsis_model" incorrect: In slot "parameters" of class "parameters": First THETA index is different than 1>