- This event has passed.
Getting Started with Bayesian PKPD Modeling Using Stan/Torsten: Practical use of Stan and R for PKPD applications
June 14 @ 1:30 pm - 4:30 pm
This is a half-day workshop on the use of Stan and R for PKPD applications held on Friday, 14 June, 13:30-17:30.
Location: Hotel C, Vasplam 4, 111 20, Stockholm, Sweden (near the PAGE 2019 meeting at the Stockholm Waterfront Congress Centre).
We will demonstrate the use of Stan (http://mc-stan.org/) and the R package rstan for Bayesian PKPD modeling. Stan is a flexible open-source software tool for Bayesian data analysis using Hamiltonian Monte Carlo (HMC) simulation—a type of MCMC simulation. Via the examples you will see how to implement nonlinear regression models, nonlinear mixed effects models and additional programming required for population PKPD models. The latter includes learning how to deal with dosing and observation event schedules specified using NONMEM conventions. Models requiring numerical solution of differential equations will also be demonstrated. We will discuss the pros and cons of Stan relative to other tools.
Who: Interested scientists with knowledge and experience in PKPD modeling, nonlinear mixed effects modeling, use of R (or S-PLUS) and a basic understanding of Bayesian principles.
$300 for industry
$150 for academia/government/non-profit
$75 for students
- Brief review of key Bayesian statistical principles
- Bayes Rule
- Bayesian modeling & inference process
- Computation for Bayesian modeling
- Key challenge of Bayesian modeling and inference: sampling from high-dimensional probability distributions
- General computational approach: posterior simulation
- Brief intro to Markov chain Monte Carlo simulation
- Stan and rstan basics
- What is it?
- How do I get it?
- How do I run it?
- Demo: Linear regression
- Introduce PK/PD modeling case study to be used throughout the workshop
- Example 1: Simple nonlinear regression
- Topics in Bayesian model development using Stan I
- R tools for running Stan and analyzing MCMC simulations
- Assessing convergence
- Programming hierarchical models (aka mixed effect or population models)
- Example 2: Nonlinear mixed effects
- Topics in Bayesian model development using Stan II
- User-defined functions
- Torsten library of PKPD functions for Stan
- Built-in models: 1 and 2 compartment models with 1st order absorption
- Numerical solution of user-specified ODEs
- Why Stan?
- vs ML tools (NONMEM, Monolix, Phoenix, nlme/r, etc.)
- vs BUGS variants (WinBUGS, OpenBUGS, JAGS)
- Parallel computation with rstan
- Example 3: Population PK.
- Additional topics & closing discussion
- Dealing with censored data in Stan, e.g., BQL data
- Example 4: Population PKPD using an indirect action model
- What can you do in Stan that you can’t do with your current tools?
- What didn’t we cover?