2.5 Population Pharmacokinetic Model Building
Based on previous experience and visual inspection, a 2-compartment
structure PK model was selected based on the objective function value
(OFV) using the log-likelihood ratio test, Bayesian information
criterion (BIC), the goodness-of-fit criteria and visual predictive
check.
Unexplained interindividual variability (IIV) in a PK parameter was
assumed to be log-normally distributed. The relationship was shown in
equation 1.
Equation 1:\(\text{Log\ }\left(\text{Pi}\right)=Log\ \left(P_{\text{pop}}\right)+\ \eta_{i}\)
where Pi is the individual parameter; Ppop represents the typical
parameter (fixed effect); \(\eta_{i}\) denotes the random effect.
Intra-individual or residual variability (RV) was modeled according to
the Monolix library. Proportional error model was chosen based on the
model fitting (Equation 2).