Understanding Infectious Disease Modelling
We want to find the most effective methods to utilize when modelling epidemics. To determine this, we are conducting interviews with leaders in the field of infectious disease modelling and studying novel ways of using models to help improve policy.
The Four Pillars of Our Work
Learning from Leaders in the Field
For this aspect of the project, we will be examining when and why different modelling methods are used, with a particular focus on infectious disease modelling utilizing COVID-19 as a case study. We are, however, investigating how modelling methods are employed in other fields and contexts as well. We are especially interested in the technical and design choices that go into building models and the associated advantages, disadvantages, and policy implications. As part of this exploration, we will also be delving into model accessibility across various stakeholders, including modelling veterans, policymakers, new users, and others trying to understand these tools and their outputs. To garner these insights we will be interviewing* and surveying** members of these diverse groups. In doing so, we hope to accomplish the following four goals:
Uncover trends in the usage of different modelling methods,
Present a clear overview of the field of infectious disease modelling,
Identify resources, especially those highlighting best practices, that can be utilized to learn about these methodologies,
And gain a greater understanding of how modelling technologies are incorporated within policy and decision-making processes.
*If you are interested in participating please contact Elise Racine at firstname.lastname@example.org.
**You can find a link to our survey here.
Combining Agent-based Models with
For this part of the project, we aim to test whether agent-based models (ABMs) can be used to improve priors relevant to optimal policy interventions. We hypothesise that ABMs are well suited to accomplish this task if heterogeneity within the population is relevant, or policy interventions are adapted according to the current state of the system. Specifically, we study the problem of repeatedly allocating a limited number of antibody tests during an epidemic, as this problem has been faced by policy-makers in the course of the COVID-19 pandemic. We investigate whether ABMs can be used to improve the priors of individual-based policies like this without requiring extra data so that they are more effective and safer when they are put into use on real people. If our method can be implemented successfully, it may be able to reduce both public health costs and lockdown costs beyond current standard practice.
Making Knowledge More Accessible
Epidemiological modelling was a well-established field long before COVID-19. With the pandemic, the discipline has seen even more attention. In an effort to make these models and methodologies more accessible, we have reviewed, summarized, and shared a number of key resources.
For readers with little to no experience or those who just want a refresher, the Common Terms section and Introductory Reading List are great places to get started. The first provides easily comprehensible definitions for the basic terminology you are likely to come across as you delve deeper into the field. The latter provides a great overview of the high-quality scholarship that has been done in the past.
For those who complete this list or are already familiar with epidemic modelling, we recommend you check out the Further Reading List, which will direct you to additional sources. If you would like to learn more about ABMs specifically, we recommend you take a look at BehaveLab's Youtube Channel for workshops and tutorials on the topic. For initiatives that can help you visualize agent-based models in practice, head to the ABM Simulation Tools page. Finally, to learn more about how epidemiological models have been utilized in response to the coronavirus pandemic, see our Existing Models and Research Groups section.
Promoting Best Practices
To advance sound science, this projects also seeks to contribute to an ontology of best practices. We try to accomplish this goal in two ways. The first is through sharing a Best Practice Guide that includes detailed information on including the responsible use of statistics, the utilization of multiple models to benefit from their individual strengths, and formal frameworks for using agent-based models in science from conception to publication. The second is by replicating and highlighting these principles in our own work.*
*Best practice principles, including transparency, have guided our research from the onset. In line with this, we will shortly be sharing more detailed documentation demonstrating how we have incorporated these standards throughout the project.
The Research Group
Project Leads: Professor Joanna Bryson, Professor Slava Jankin
With: Dr. William Lowe
Research Associates: Elise Racine, Jonathan Barnes-Nunn, Daniel Privitera, Philipp Jäger
In cooperation with the Hertie School Data Science Lab