Goodness-of-fit and model evaluation
The goodness-of-fit plots for both PCr and Cr observations of the final
model are shown in Figure 2 . The plots for CWRES (PCr and Cr)
versus time or population predictions are shown in Figures 2A and 2B
which were used to detect any misspecifications in the model. The CWRES
of the final pharmacokinetic model distributed around the line y = 0
indicating no apparent systematic bias were observed. Scatter plots of
observation (PCr and Cr concentrations) versus population or individual
predictions are shown in Figures 2C and 2D, and the solid lines are the
unity. The data points distributed evenly around the line of identity in
the final model (Figure2C). These diagnostic plots suggested a good fit
of the proposed final model to the data.
The evaluation of 1000 bootstrap runs with a success rate 98.3% (983
out of 1000 resampling datasets were successful in optimization) showed
an acceptable stability of the final population pharmacokinetic model.
Bootstrap results of median parameter values and 95% CIs are listed inTable 2 . The median values are similar to the final parameter
estimates and 95% CIs are closed to the values obtained during final
data fitting. The VPC results (PCr and Cr) are shown in Figure
3 . In the VPC plot, 90% prediction interval (90% PI) is the area
between predicted 5th and 95th percentiles (black dashed lines) and the
majority of actual observations fell within the 90% PI. The predicted
5th 50th and 95th percentiles (black lines) are closed with the observed
5th 50th and 95th percentiles (red lines), respectively. This plot
indicated an adequate predictive ability of the final population model.