Modelagem probabilística da dinâmica da Zika usando modelos hierárquicos bayesianos

• Autor(a) principal: Bastos, Marcio Maciel
• Data de Publicação: 2018
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional do FGV (FGV Repositório Digital)
• Texto Completo: http://hdl.handle.net/10438/21991
• topic: Zika virus; Estimativa da epidemia de Zika; Matemática; Vírus da Zika; Modelagem de dados; Teoria bayesiana de decisão estatística; Epidemiologia - Modelos matemáticos

The Zika virus (ZIKV) is a pathogen of the family Flaviviridae, transmitted in Brazil mainly by the mosquito Aedes aegypti and in less extent by sexual relations. In addition to symptoms common to dengue and chikungunya, the zika virus is also capable of causing irreversible damage to the nervous system, in adults it is related to Guillain-Barr´e syndrome and in fetuses it causes microcephaly. The Health Department of Rio de Janeiro maintains a database with records of patients who sought care and was infeccted with Zika. Our study seeks to estimate the true size of the epidemic that occurred in the year 2016 and the parameters that fit to explain the dissemination process. To make these estimates, we used the data provided by the Health Department and a hierarchical Bayesian model adapted to the SIR epidemiological model. We perform the inference process through modern sampling techniques such as Automatic Differentiation Variational Inference (ADVI), Stein Variational Gradient Descent (SVGD) and No-U-Turn (NUTS).