Monte Carlo simulation of classical swine fever epidemics and control. II. Validation of the model.

A stochastic and spatial simulation model was developed to simulate the spread of classical swine fever virus among herds in a certain area. A model is a simplification of a real system. The mechanisms and parameters are often not exactly known. Validation is necessary to gain insight into model behaviour and to identify risk factors with great impact on the response variables. Several risk factors such as incubation period, number of daily farm contacts, probability of detection, probability of infection after contact, probability of local spread and time from infection to infectivity were considered in the model as probability distributions in order to take the stochastic component of disease dynamics into account. In order to estimate the effects of the risk factors on the response variables mean size and duration of epidemics, a sensitivity analysis was performed. A fractional factorial design with two-level factors (2(7-2) design) was developed to gain the maximum strength with minimum demand on the calculating capacity. The main factors were unconfounded with any other main factor and also unconfounded with two-factor interactions. Apart from the time from infection to infectivity, all risk factors had a significant effect on the mean size and duration of epidemics (p<0.05). Eight two-factor interactions had a significant influence as well (p<0.05). Mainly, two-factor interactions with probability of detection were significant thus emphasising the impact of a rapid detection of outbreaks. The reaction of the simulation responses to changing of the parameter values was consistent with the expected reaction.