Model diagnostics and selection (Ep. 11)

This episode is a discussion on model evaluation and selection. I discussed diagnostic plots using predictions and residuals. I covered approaches for model selection and factors to consider. The focus of this episode is on models that predict continuous endpoints like drug concentration, biomarkers, or other continuous endpoints. Several figures are mentioned in the podcast and links to those images can be found below.

Links discussed in the show:


Figure 1: Observed vs Predicted for single-subject model (example of a good model fit). Figure generated using ggplot in R.
Figure 2: Observed vs Predicted for single-subject model (example of a poor model fit) using logarithmic scales. Figure generated using ggplot in R.
Figure 3: Observed vs Population and Individual Predicted for population model. Figures generated using ggCertara package in R.
Figure 4: Weighted residuals versus time, time after dose, and population prediction for population model. Figures generated using ggCertara package in R.
You can connect with me on LinkedIn and send me a message
Do you need help with model selection? Send me a message
Sign up for my newsletter


Copyright Teuscher Solutions LLC

All Rights Reserved