DECOM - Departamento de Computação

URI Permanente desta comunidade

http://www.repositorio.ufop.br/handle/123456789/575

Navegar

Navegando DECOM - Departamento de Computação por Assunto "Aedes aegypti"

Agora exibindo 1 - 3 de 3
  • Nenhuma Miniatura disponível
    Item
    Aquisição de imagens digitais e identificação dos ovos do mosquito Aedes Aegypti baseado em um modelo de aprendizado profundo.
    (2019) Garcia, Pedro Saint Clair; Cámara Chávez, Guillermo; Cámara Chávez, Guillermo; Ferreira, Anderson Almeida; Bianchi, Andrea Gomes Campos; Saúde, André Vital
    O mosquito Aedes aegypti pode transmitir algumas doenças, o que faz o estudo da proliferação deste vetor uma tarefa necessária. Com o uso de armadilhas feitas em laboratório, denominadas ovitrampas, é possível mapear a deposição de ovos numa determinada comunidade. Uma máquina fotográfica acoplada a uma lupa foi utilizada para adquirir imagens contendo os elementos (ovos) a serem contados. Essas imagens foram processadas a partir de um sistema de cores com o objetivo de encontrar a cor negra, que corresponde `a cor dos ovos. A partir dessas imagens já trabalhadas, foi realizado um processo de transferência de aprendizado com uma rede neural convolucional (CNN). A intenção era separar os elementos que realmente eram ovos dos demais. Por meio desse método, foi possível identificar cada ovo como um simples objeto. Em 90% das imagens testadas a contagem realizada pelo modelo em relação ao número real de ovos foi considerada de correlação perfeita. Para as demais 10% das imagens de teste, a contagem foi considerada de forte correlação, isso aconteceu em imagens que continham uma alta densidade de ovos ou que continham elementos negros que se pareciam com ovos do mosquito.
  • Nenhuma Miniatura disponível
    Item
    Multiscale analysis and modelling of Aedes aegypti population spatial dynamics.
    (2011) Lana, Raquel Martins; Carneiro, Tiago Garcia de Senna; Rocha, Nildimar Honorio; Codeço, Cláudia Torres
    Population dynamic models requires the evaluation of the best scale of analysis. This work analyses three spatial scales in the context of the mosquito Aedes aegypti, main vector of dengue fever. One scale is the neighborhood, the others scales are the census tract and the lot. A geographical database was developed including point maps with trap locations, number of eggs collected per trap per week, polygons of census tracts, census data, among others. For simulation purposes, a layer of regular cells (10 x 10 meters) was created to store the model’s inputs and outputs. A population dynamic model with temperature as input variable was parameterized and fitted to the neighborhood and census tract data. For the lot level, an allocation procedure was developed as the spatial resolution was higher than the data resolution. This procedure couples the population dynamic model with a kernel density map. Results indicate that at the neighborhood level, the population model captured well the overall pattern with lower mosquito density during the cold season and larger during the warm season. However, in the first warm season, two peaks did not fit well, suggesting the importance of investigating the role of other variables in the dynamics of Aedes aegypti. At the census tract level, we found no evidence of spatial clustering. At the lot level, the allocation model represented well the overall summer to winter variation in hotspot intensity. The cost of vector surveillance is high and the procedures proposed here can be used to design optimized control strategies and activities.
  • Nenhuma Miniatura disponível
    Item
    Seasonal and nonseasonal dynamics of Aedes aegypti in Rio de Janeiro, Brazil : fitting mathematical models to trap data.
    (2014) Lana, Raquel Martins; Carneiro, Tiago Garcia de Senna; Rocha, Nildimar Honorio; Codeço, Cláudia Torres
    Mathematical models suggest that seasonal transmission and temporary cross-immunity betweenserotypes can determine the characteristic multi-year dynamics of dengue fever. Seasonal transmis-sion is attributed to the effect of climate on mosquito abundance and within host virus dynamics. In thisstudy, we validate a set of temperature and density dependent entomological models that are built-incomponents of most dengue models by fitting them to time series of ovitrap data from three distinctneighborhoods in Rio de Janeiro, Brazil. The results indicate that neighborhoods differ in the strength ofthe seasonal component and that commonly used models tend to assume more seasonal structure thanfound in data. Future dengue models should investigate the impact of heterogeneous levels of seasonalityon dengue dynamics as it may affect virus maintenance from year to year, as well as the risk of diseaseoutbreaks.