An international modelling challenge to predict and control a virtual African swine fever (ASF) epidemic at the pig-boar interface

Context and issues

Organising a modelling challenge creates a unique and challenging environment to improve the ability of modellers to advise policy makers in a timely manner. Such challenges are organised in human health, for example for influenza, dengue, Ebola. However, no such event has yet concerned the animal health epidemiology community. This first animal epidemiology modelling challenge will help to improve the preparedness of the international teams involved when faced with emerging threats such as African swine fever (ASF). Such a challenge will also promote international collaborations and increase the visibility of INRAE.

Results

During the first year of the project, the challenge's organizers (P. Ezanno, M. Mancini and S. Picault (BIOEPAR, Nantes), and T. Vergne (IHAP, Toulouse)) developed an original spatio-temporal model of ASF virus circulation at the interface between wildlife (wild boar) and animal husbandry (industrial or free-range) in a European context. This model integrates the regulatory management methods as well as possible alternatives. The organising team was supported by a group of experts composed of E. Gilot-Fromont (VetAgroSup), E. Baubet, S. Rossi & E. Marboutin (OFB), and C. Peroz & C. Belloc (Oniris), who have expertise in the epidemiology of ASF, regulated ASF control in the animal husbandry sector and regulated ASF control in the wildlife sector. The model constructed was able to generate synthetic data similar to that which might be available during a real epidemic.

For five months from 28/08/2020, the launch date, the nine international teams participating in the challenge are using these simulated synthetic data in real time to predict the outcome of the epidemic. The data is progressively made available to the teams to mimic the progress of the outbreak. Participants attempt to answer management questions, such as identifying the best control measures to implement or estimating the probability of a second wave of the epidemic.

Participants can use any type of model, statistical or mechanistic. Indeed, previous modelling challenges in human health have shown that simple statistical models can be more useful in certain situations than complex mechanistic models. On the other hand, mechanistic models are often very useful for comparing intervention strategies. Comparing this diversity of approaches will make it possible to compare the predictive capacities of these methods and to identify the points of progress for a better reactivity of the modellers in the event of a real crisis.

Website: https://www6.inrae.fr/asfchallenge/

Twitter account: @AsfMod

Perspectives

This project provides an inspiring platform for the exchange of knowledge and expertise on animal health modelling at the livestock-wildlife interface. It will also bring together the community of animal health modelling epidemiologists to improve our collective capacity to respond to a crisis situation. Finally, it will be an opportunity to publish a special issue in a high-impact journal on the collective experience and the results obtained.