I- How can I generate sparse PK data set (drug plasma concentration data) using the following model and parameters:

• One-compartment model with bolus dosing and first-order elimination as follows: fi(Xi,Bi)=(D/Bi2)exp(−Bi1tij/Bi2),j=1,...,ni

where Bi1 = Clearance ; Bi2 = volume of distribution

• The administered dose D at time 0 (D=25.7μg, mean of the initial dose of drug for all individuals

• Volume of the central compartment (VCentral) in liter

V∼logN(0.00613,0.0021002)

where the volume of distribution will simulated according to a log-normal distribution.

• Clearance parameter (Cl_Central) in liter/hour

CL∼0.7logN(0.007345,0.0010902)+0.3logN(0.003295,0.006782)

where the clearance (Cl) was simulated a mixture of log-normal random variables, with a minor (30 %) and a major (70 %) sub-population.

• The clearance and the volume of distribution will simulated independently: cov(Cl,V)=0

• The measurement error is assumed to be proportional to the expected concentrations:

g_i(X_i,B_i) = µ* diag(f_i(X_i,B_i))

where μ* is the only fixed effect (p = 1). In the simulations, µ* will be set equal to 0.128

• For sparse datasets (N = 25 sample size of individuals), I would like to simulate: 2 observations per individual for 25% of the individuals: one early and one late, 3 observations per individual for 15% of the individuals: one early, one intermediate and one late, 1 observation per individual for 60% of the individuals: any randomly position of three.

• The observation times is: 12 minutes (min) for the early observation, 6 hours (h) for the intermediate one and 246 h for the late one.

II- If I have these data (dependante variable = drug plasma concentration; independante variables= ID, Time, Dose =25.7μg mean of the initial dose of drug for all ) and

I want to estimate population PK parameters (Cl and V) using NPAG method, what are the different steps I must do? But I have installed Pmetrics Package and gfotran compiler with PMetrics mannuel, I have some difficulties to understand well when i read and implemented some exemple whose are whitin. Perhaps that I translate bad the content, I use french like work language.