Decision Modelling in Python

An introduction to the MISCore Discrete-Event Microsimulation Package to Simulate Disease and Intervention based on MISCAN Models

28 June 2026, 8:35 AM – 12:00 PM (local time)

Faculty

  • Luuk van Duuren – Doctoral Student, Department of Public Health, Erasmus University Medical Center, Netherlands
  • Chiara Brück, PhD – Postdoctoral Researcher, Department of Public Health, Erasmus University Medical Center, Netherlands
  • Duco Mülder – Early-Career Scientist, International Agency for Research on Cancer, Lyon, France
  • Koen de Nijs – PhD Candidate, Department of Public Health, Erasmus University Medical Center, Netherlands
  • Erik Jansen, PhD – Postdoctoral Researcher, Department of Public Health, Erasmus University Medical Center, Netherlands

Course Overview

Microsimulation models are powerful tools to inform decision making in healthcare. Compared to randomized controlled trials, they can evaluate multiple interventions in less time, at lower cost, and while accounting for uncertainty. These models simulate individuals one-by-one and, in contrast to Markov models, allow for heterogeneity between individuals to capture complex disease patterns. They are increasingly implemented in scientific programming languages such as R and Python due to their flexibility, computational power, and transparency.

All microsimulation models require core functionalities such as event scheduling, output logging, model calibration, and common random number techniques. However, few frameworks provide a comprehensive, ready-to-use set of these features. As a result, modelers often develop these components independently or adapt existing code, leading to duplication of effort and non-standardized, unvalidated implementations.

This course introduces the MISCore (MISCAN Core) Python package, which provides ready-to-use core functionalities for health economic modelling. In addition to core simulation features, MISCore includes built-in tools for cost-effectiveness analysis, probabilistic sensitivity analysis, and output stratification. MISCore was developed as the simulation framework for the MISCAN (Microsimulation Screening Analysis) family of models, which have a 40-year history of informing screening policy, including contributions to USPSTF guidelines for colorectal, cervical, and lung cancer screening, as well as policy development across Europe. While originally developed for MISCAN models, MISCore is broadly applicable to a wide range of modelling contexts and is being prepared for public release as a non-commercial Python package.

The course follows the three phases of simulation modelling: model application, model analysis, and model development. Each phase begins with a short conceptual introduction followed by hands-on Python exercises.

Participants will begin with the application phase, learning the general structure of MISCore models, running simulations using a disease model, and evaluating intervention strategies. In the analysis phase, participants will work with model outputs and conduct cost-effectiveness analyses using MISCore functionality. In the development phase, participants will modify a simple example disease model to understand the underlying code structure and gain the skills needed to begin building their own models. The course concludes with a brief overview of additional MISCore features, documentation resources, and licensing considerations.

By the end of the course, participants will be familiar with discrete-event microsimulation modelling in Python using MISCore and will be equipped to continue developing their skills through available tutorials and documentation.

Learning Objectives

Participants will learn how to:

  • Familiarize themselves with the key features of MISCore for developing, applying, and analyzing microsimulation models in Python
  • Use an existing disease model, such as MISCAN-Endometrium, and simulate intervention strategies
  • Analyze model outputs and perform cost-effectiveness analyses using MISCore functions
  • Understand the underlying code structure of MISCore models to support development of new models

Course Format

This course combines short lectures with hands-on Python exercises structured across three phases:

  • Model Application: Set up and run simulations using a MISCore disease model and evaluate intervention strategies
  • Model Analysis: Analyze outputs and perform cost-effectiveness analyses using MISCore tools
  • Model Development: Modify a simple model to understand code structure and begin building new models

The session also includes a brief introduction to discrete-event microsimulation modelling, a scheduled break, and a concluding overview of additional MISCore features and licensing.

Participant Requirements

Participants should be familiar with a scientific programming language such as Python, R, or a comparable language.

Instructions for installing Python, PyCharm, and the MISCore package will be provided prior to the course. Participants are expected to bring a laptop with the required software installed.

A basic understanding of decision modelling and cost-effectiveness analysis is helpful but not required.