Bayesian Benchmarks

With Bayesian methods becoming more popular in the pharmacometrics community, a thorough comparison of the available software is necessary to determine the accuracy and (speed) performance of the three main software packages that implement Bayesian methods for pharmacometrics - NONMEM, Pumas, and Stan/Torsten.

This repo contains code for all 3 of the above-mentioned software to implement Bayesian estimation for some commonly used models:

• A one-compartment model with IV infusion and first-order elimination (Gabrielsson and Weiner, 2000)
• A one-compartment model with first-order absorption and first-order elimination (Gabrielsson and Weiner, 2000)
• A two-compartment model with first-order absorption and first-order elimination (Gabrielsson and Weiner, 2000)
• A two-compartment model with first-order absorption and Michaelis-Menten elimination (Gabrielsson and Weiner, 2000)
• A two-compartment PK model with IV infusion and a Friberg-Karlsson PD model (see Friberg 2002)

The code in this repository is meant to be as minimal as possible to compare performance. For a more thorough workflow that one might use in practice see here for NONMEM, here for Pumas, and here for Stan/Torsten.