Model construction and sensitivity analysis
The risk assessment models were constructed in R (R-language, version 3.4.4). For the Monte Carlo simulation, the analysis for each country was repeated 1,000 times. The probability of introduction of each country was clustered using the Jenks natural breaks optimization and mapped using the same program.
Sensitivity analysis for both pathways was performed to evaluate the influence of the changes in the value of the input parameters on the model outcomes. A regression coefficient (βi) was calculated between each input and the probability of ASF introduction by month. The inputs that were most influential were determined and analyzed using spider graphs, which shows the impact that changes in the input parameters from a minimum of a 50% reduction to a maximum of a 50% increase have on the final probability using a linear regression approach. All analyses were conducted in R.