Diagnostics and Model Evaluation

Evaluating model fit, assumptions, and reliability before making decisions

A practical framework for evaluating pharmacometric models using residuals, simulation-based diagnostics, and model comparison.

Welcome

In pharmacometrics, building a model is only part of the process.

The critical question is:

Can we trust this model?

This module focuses on how to evaluate models before using them for decisions.

You will learn how to:

  • assess model fit
  • detect problems and misspecification
  • compare competing models
  • determine whether a model is reliable

Why this module matters

Models are used to support:

  • dose selection
  • trial design
  • regulatory decisions

But:

A model that fits data well is not necessarily a good model.

Without proper evaluation:

  • incorrect assumptions may go unnoticed
  • predictions may be unreliable
  • decisions may be flawed
Warning

Model evaluation is not optional—it is essential for making trustworthy decisions.

Learning objectives

By the end of this module, you will be able to:

  • Explain why model evaluation is necessary
  • Interpret residuals and goodness-of-fit plots
  • Use simulation-based diagnostics (VPC/PPC)
  • Compare and select models appropriately
  • Connect model evaluation to decision-making

Course structure

This module builds a complete evaluation workflow:

  1. Why Model Evaluation Matters
    • Why fit is not enough
    • The role of assumptions
  2. Residuals and Goodness-of-Fit
    • Observed vs predicted
    • Residual diagnostics
  3. Simulation-Based Diagnostics (VPC/PPC)
    • Distribution-level evaluation
    • Variability and uncertainty
  4. Model Comparison and Selection
    • AIC, BIC, likelihood
    • Balancing fit and complexity

Key idea

Model evaluation answers:

Is this model reliable for the question we care about?

It is not about:

  • perfect fit
  • maximizing likelihood

It is about:

  • understanding limitations
  • ensuring robustness
  • supporting decisions

What you’ll be able to do after this module

  • Identify when a model is unreliable
  • Diagnose common modeling problems
  • Compare models effectively
  • Select models that support decisions
  • Avoid common pitfalls in model evaluation

Get started

Begin with Why Model Evaluation Matters to understand the role of evaluation in the modeling workflow.