Introduction to Bayesian PK/PD Modeling and the New BUGS PK/PD Model Library

Date: Sunday, October 4, 2009

Registration fee:  Non-student ($500), Student/fellow ($50)

Maximum class size: 50

Sponsor: Metrum Institute

Instructors/Teaching Assistants: Bill Gillespie, Marc Gastonguay, Matthew Riggs, Jim Rogers, Bill Knebel, John Mondick

Workshop summary:  This will be an abbreviated version of our MI250 course, designed to provide a one-day overview of the content, with scripts and code for hands-on examples provided for attendees.

The workshop will provide an introduction to Bayesian modeling theory and the practical use of WinBUGS and R for PK-PD applications. In addition to basic concepts, this new course includes instruction on the BUGS PK-PD model library, an open-source tool developed by Metrum Institute. This library facilitates the implementation of population PK-PD models in WinBUGS for compartmental models described by algebraic, general linear and differential equations. Hands-on software requirements include WinBUGS and R (both freely available). Students will provide their own laptop computers.

Suggested Prerequisites: Experience with PK-PD modeling and some familiarity with nonlinear regression, mixed effects modeling and R (or S-Plus) would be helpful. Applicable MI courses include: MI 210 and MI212.

Schedule:

8:00 - 10:30 am

  • Introduction.
  • Current & potential of Bayesian methods in clinical drug development
  • Introduction to Bayesian statistical principles & methods
  • Computation for Bayesian Modeling
  • Key challenge of Bayesian modeling & inference
  • General computational approach: posterior simulation
  • Brief introduction to Markov chain Monte Carlo simulation
  • WinBUGS Basics.
  • What is it, how does it work, and where do I get it?
  • WinBUGS demo: linear regression
  • Using the WinBUGS model language
  • Entering data and initital estimates
  • Generating MCMC samples
  • Analyzing and interpreting the results
  • Hands-on Session 1: Simple Nonlinear Regression

10:15 - 10:30  Morning Break

10:30 am - 12:00 pm

  • Topics in Bayesian Modeling Development using WinBUGS I
  • Using WinBUGS scripts
  • R tools for running WinBUGS and analyzing MCMC simulations.
  • Assessing convergence
  • Topics in Bayesian Modeling Development using WinBUGS II
  • Programming hierarchical models
  • Hands-on Session 2: Nonlinear Mixed Effects Model

12:00 - 1:00 pm  Lunch

1:00 - 2:30 pm

  • Topics in Bayesian Modeling Development using WinBUGS III
  • Model evaluation and comparison
  • Dealing with censored data in WinBUGS
  • PK-PD modeling using BUGSModelLibrary built-in models
  • Hands-on Session 3: Population PK

2:30 - 2:45 pm  Afternoon Break

2:45 - 4:30 pm

  • Topics in Bayesian Modeling Development using WinBUGS IV
  • Informative prior distributions in clinical pharmacology applications
  • Assessing sensitivity to prior distributions
  • Models based on linear and nonlinear ODE's using BUGSModelLibrary
  • Using the cut function in WinBUGS
  • Hands-on Session 4: Simultaneous PK-PD Modeling Using an Indirect Action Model
  • Additional Topics & Closing Discussion.
  • Selected published examples of Bayesian applications
  • Considerations in deciding whether to use Bayesian modeling with WinBUGS versus maximum likelihood modeling with NONMEM (or other ML tools)
  • What didn't we cover?

4:30 PM Workshop Concludes