Graph theory in veterinary epidemiology : modelling an outbreak of classical swine fever

Abstract: One objective of this study has been to demonstrate how graph theory can be a useful tool for intervention in the case of an outbreak of a contagious disease. By using graph theory algorithms on data from the Swedish surveillance network system (Grisregistret) and combining this with the power of graph layout engines for visualization of interherd movements, valuable information can rapidly be provided from data that is easily available. Thus a clear line of priority in the strategy of fighting the outbreak can be established from day 0. This can be invaluable during the first couple of days, before more thorough information concerning all different types of contacts has been collected by veterinarians during farm visits. Thus rapid control measures can be implemented and this should increase the likelihood that possible infectious holdings are found during an early stage of the outbreak. Combining the information of the geographical location of an infectious holding and the interherd movements during the relevant time period on the same map can improve the understanding of the spread of the outbreak. It would also be possible to add information concerning other types of contacts, when such information becomes available, to make the picture clearer. Another objective has been to model a putative outbreak of classical swine fever without any interventions from authorities. The epidemiological model uses data from the Swedish surveillance network system of pig movements, Grisregistret, based on the reported flows of pigs between farms. The model includes all interherd pig movements that occurred within a community in southern Sweden during 6 months. Both a high virulent strain and a low virulent strain of classical swine fever virus have been modelled. The epidemiological model is based on a SEIRD model for each holding, which have been interconnected to each other through the reported movements of pigs.