Overview

Description

Draw population parameters using the covariance matrix of the estimates.

Usage

Arguments

n
the number of vectors of population parameters (default = 1)
project
a Monolix project, assuming that the Fisher information Matrix was estimated by Monolix
fim
the Fisher Information Matrix estimated by Monolix. fim=“sa”, “lin” (default=“sa”)
parameter
a data frame with a column pop.param (no default), a column sd (no default), and possibly a column trans (default =‘N’). Only when project is not used
corr
correlation matrix of the population parameters (default = identity). Only when project is not used
kw.max
maximum number of trials for generating a positive definite covariance matrix (default = 100)


Examples

Using a Monolix project

We use the Monolix project warfarinPK1_project.

In this project, the Fisher Information Matrix was estimated both by linearization and by stochastic approximation. By default, simpopmlx uses the FIM estimated by stochastic approximation (sa).

## [INFO] The lixoftConnectors package has been successfully initialized:
## lixoftConnectors package version -> 2019.1
## Lixoft softwares suite version   -> 2019R1
##   pop ka_pop V_pop Cl_pop omega_ka omega_V omega_Cl     a     b
## 1   1  0.660  7.66  0.138    0.905   0.204    0.253 0.595 0.138
## 2   2  0.760  7.46  0.131    0.852   0.273    0.303 0.647 0.111
## 3   3  0.461  8.07  0.142    0.801   0.240    0.300 0.638 0.134

the FIM estimated by linearization can be used by setting fim="lin"

##   pop ka_pop V_pop Cl_pop omega_ka omega_V omega_Cl     a      b
## 1   1  0.920  7.14  0.131    1.347   0.190    0.284 0.857 0.0835
## 2   2  0.473  7.00  0.148    0.768   0.249    0.319 0.793 0.1090
## 3   3  0.699  7.82  0.137    1.216   0.193    0.331 0.716 0.1086

In the project warfarinPK2_project, \(ka_{\rm ppop}=0.5\) (fixed) and a correlation between \(\eta_{ka}\) and \(\eta_V\) is estimated.

##   pop ka_pop V_pop Cl_pop omega_ka omega_V omega_Cl corr_V_Cl     a      b
## 1   1    0.5  8.04  0.140    0.677   0.217    0.257     0.324 0.617 0.0732
## 2   2    0.5  7.97  0.140    0.619   0.242    0.355     0.398 0.573 0.0747
## 3   3    0.5  7.47  0.123    0.915   0.223    0.287     0.422 0.686 0.0563

In the project warfarinPK3_project, \(\omega_{ka}=0\) (fixed) and the weight is put both on \(V\) and \(Cl\).

##   pop ka_pop V_pop beta_V_lw70 Cl_pop beta_Cl_lw70 omega_ka omega_Cl     a
## 1   1    0.5  7.39       0.770  0.142        0.361    0.805    0.245 0.508
## 2   2    0.5  7.90       0.702  0.138        0.531    0.783    0.246 0.474
## 3   3    0.5  7.64       0.885  0.143        0.775    0.964    0.312 0.517
##       b
## 1 0.116
## 2 0.112
## 3 0.101


Using a data frame

In this example, the three first parameters and the seventh one are normally distributed, the three other ones are log-normally distributed. Then, pop.param are the medians of these distributions while sd are the strandard deviations of the normal distributions.

##   pop    1     2      3     4     5     6     7
## 1   1 1.66 0.529 0.0218 0.495 0.156 0.213 0.670
## 2   2 1.67 0.565 0.0214 0.366 0.164 0.194 0.667
## 3   3 1.40 0.426 0.0203 0.438 0.135 0.191 0.622
##   pop    1     2    3     4     5     6   7
## 1   1 1.44 0.357 0.02 0.418 0.122 0.225 0.7
## 2   2 1.31 0.475 0.02 0.401 0.134 0.226 0.7
## 3   3 1.38 0.404 0.02 0.456 0.135 0.201 0.7

Logit-normal distributions can be used with the alias ‘G’. By default, the support of a logit-normal distribution is (0,1).

##   pop    1     2      3     4     5     6     7
## 1   1 1.65 0.915 0.0244 0.425 0.143 0.199 0.757
## 2   2 1.50 0.443 0.0169 0.486 0.130 0.207 0.671
## 3   3 1.62 0.407 0.0170 0.485 0.140 0.197 0.785

additional columns lim.a and lim.b should be added in order to define a logit-normal distribution on interval (a, b)

##   pop    1     2      3     4     5     6     7
## 1   1 1.59 0.253 0.0163 0.381 0.173 0.187 0.650
## 2   2 1.32 0.496 0.0164 0.432 0.167 0.191 0.778
## 3   3 1.70 0.484 0.0242 0.438 0.129 0.204 0.709