Evaluation and validation
The addition of covariates greatly
improved the stability of model, as a result, the OFV decreased from
1774.00 (base model) to 1673.89 (final model). The observation versus
population and individual predictions (Figure 6A, B) were generally
evenly distributed around the identity line, except for some positive
bias at high concentrations, indicating a good model fitting. Of the
conditional weighted residuals (CWRES) versus time after dose and
population predictions (Figure 6C, D), the CWRES were symmetrically
distributed around y=0 line and mostly within ±2 unit, confirming that
the model has a good accuracy. The population typical value of final
model closely agreed with mean estimated from bootstrap and is within
95% CI, which demonstrate that the model prediction is accurate and
stability (Table 2). According to VPC (Figure 7), the major observations
was distributed within 95% CI of simulated values, indicating a good
predictive performance of final model.