Análise das redes de trânsito animal integrada à simulação da difusão de enfermidades infecciosas

• Ano de defesa: 2011
• Autor(a) principal: Carolina Silva Pena
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: medidas de centralidade; Modelos SIR; redes dinâmicas; Estatística; Animais tráfego; Teoria da estimativa
• Link de acesso: http://hdl.handle.net/1843/ICED-8FUGNS

The animal's transit in great part of Brazilian territory is electronically documented by Guides of Animal's Transit (GTAs). In this essay, we developed an application called EpiGTA, which is able to simulate the transmission of contagious animal diseases from the traffic data provided by these guides. To simulate the transmission of an animaldisease, three types of epidemiological SIR (Susceptible - Infected - Recovered) models were used. The simplest model adopted, called the Farm Level (Vernon and Keeling, 2008), treats the farm as a basic epidemiological unit. The two other ones, Greenwood (Greenwood, 1931) and Reed-Frost (Abbey, 1952), simulate the transmission ofdiseases using the animal as the basic epidemiological unit.The incorporation of Greenwood and Reed-Frost models in network structures allows to assign weights to the connections between each pair of vertices (farms), according to the flow of animals and the number of infected ones in the herd of its farm of origin. The implementation discussed through the text focuses the spreading of foot and mouthdisease in Mato Grosso do Sul. The simulations showed that the estimates of the average daily number of infected farms are very similar in the models (Greenwood and Reed-Frost), while the model Farm Level presents lower estimated values, especially with the passage of time since the epidemic beginning. On the other hand, when it comes to the average daily number of infected animals, the estimates generated by themodels Reed-Frost and Greenwood are not as close as in the estimation made for the farms. The Reed-Frost's simulations were used to estimate the vulnerability of each farm to infection by FMD. The results were compared to measures of network centrality indegree, betweenness, static neighborhood and dynamic neighborhood, and also with theflow of animals received by each vertex. The analysis has revealed that vertices with a larger dynamic neighborhood are also more vulnerable to infection by FMD.