Stochastic modelling of African swine fever in wild boar and domestic pigs: Epidemic forecasting and comparison of disease management strategies

A multi-host mechanistic model of African swine fever emergence and control in Romania

“…Beyond only direct transmission, a scoping review of articles on wild-domestic ASFV transmission concluded that when such between-host transmission occurs, indirect routes may in fact be more common (38). Our results are in accordance with one of the models out of the ASF Modelling Challenge, where preventive culling of domestic pig herds nearby to detected wild boar cases resulted in a decrease of almost 20% in the final number of infected pig herds (39). Acceptance of such a strategy by farmers could be difficult without targeted educational programs.…”

The African swine fever modelling challenge: Model comparison and lessons learnt

“…In all but one team, infection status in pig farms was defined at farm scale. The UK team however defined the health status of individual pigs within farms (Dankwa et al, 2022). Finally, all teams modelled both the temporal and the spatial infection dynamics.…”

“…With regards to virus transmission, heterogeneity arose in the choices made by teams for modelling transmission between wild boar and pig farms: the CIRAD team did not link their models for the two host species and simply modelled risk in pig farms as a function of smoothed wild boar case data (Muñoz et al, 2022); the UK (Dankwa et al, 2022) and the WUR teams modelled the force of infection from wild boar to pig farms but not from pig farms to wild boar as they found no strong evidence, based on the synthetic data provided, of transmission from pig farms to wild boar; finally, the INRAE (Beaunée et al, 2022) and Massey (Han and Vignes, 2022) teams defined a force of infection in both directions, from wild boar to pig farms and from pig farms to wild boar. Assuming no transmission from pigs to wild boar (three teams) implies that control measures targeting pig farms had, by definition, no impact on wild boar.…”

“…They fixed control measures for the duration of the prediction period, assuming measures remained similar to those of the observed period, i.e., for phase 1: regulatory measures only; for phase 2: fences installed 60 days after the detection of the first case and intensive hunting within the fenced area and a 15-km buffer around it. See teams' papers for more details (Muñoz et al, 2022;Beaunée et al, 2022;Dankwa et al, 2022;Han and Vignes, 2022). To compare their predictions to the dynamics that could have been observed with model M0 under these conditions, we ran 500 stochastic repetitions keeping control measures similar to just before the prediction period (i.e., measures already implemented on day 50 for predictions during phase 1, and on day 80 for predictions during phase 2; Fig.…”

“…The five teams that succeeded in completing the competition were asked to develop a model to fit the synthetic data at three different stages of the epidemic, to predict the spatio-temporal development of the epidemic and to assess the effectiveness of a limited number of management strategies. The approaches developed by four of the five teams are described in dedicated papers of this special issue (Muñoz et al, 2022;Beaunée et al, 2022;Dankwa et al, 2022;Han and Vignes, 2022).…”