Advanced Use of Stan, rstan, and Torsten for Pharmacometric Applications
May 29 @ 8:30 am - 4:30 pm
Montreax, Switzerland (exact location TBD)
Instructor: Bill Gillespie, Ph.D.
Metrum Research Group will present a one-day workshop entitled “Advanced Use of Stan, rstan and Torsten for Pharmacometric Applications” on Tuesday 29 May 2018. We will provide a guided hands-on experience in the advanced use of Stan, rstan, and Torsten, 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. Torsten is a Stan extension containing a library of functions to simplify implementation of PKPD models. This workshop builds on the foundations presented in our previous introductory Stan workshops. Topics include model evaluation and comparison, models with systems of ODEs, optimizing Stan code, using MCMC results for population and trial simulations, and more. You will execute Bayesian data analysis examples using Stan. Via the examples you will learn to implement population PKPD models including those involving censoring, numerical solution of ODEs, and user-defined probability distributions and likelihoods.
Full Price: $600
Academia/government attendees: $300
Materials Provided by Metrum: Metworx user login (with 1-week access post-workshop), course slides, data and code for all examples.
Requirements for Attendees: Knowledge and experience in PKPD modeling and simulation, nonlinear mixed effects modeling and the use of R (or S-PLUS).
Participants should bring a laptop that meets the following requirements:
– Browser: Chrome, Firefox, or Safari
– Internet access via Wi-Fi
Brief review of the use of Stan and rstan
– User-defined functions
– Implementing popPKPD models
– Modeling workflow using rstan
Models with systems of ODEs
– Linear case
– General case
Torsten: Prototype library of PKPD functions for Stan
– Built-in models: 1 and 2 compartment models with 1st order absorption
– Numerical solution of user-specified ODEs
Hands-on example 1: popPK using a 2 compartment model with 1st order absorption
Review of HMC/NUTS
Diagnosing and remedying sampling problems encountered w/Stan
Optimizing Stan code
– Parameterization, e.g., centered vs non-centered parameterizations for hierarchical models
– Constructing/choosing prior distributions
Hands-on example 2: popPKPD using a model based on a linear system of ODEs
Model evaluation and comparison
Use of informative prior distributions in pharmacometrics applications
Using MCMC results for statistical inference
– Population and trial simulations based on MCMC results
Hands-on example 3: physiologically-based popPKPD model
User-defined probability distributions and likelihoods
Hands-on example 4: parametric time-to-event model
What didn’t we cover?