CMStatistics 2021: Start Registration
View Submission - CMStatistics
B1141
Title: A latent spatial model for pandemic prediction Authors:  Dani Gamerman - Universidade Federal do Rio de Janeiro (Brazil) [presenting]
Marcos Prates - Universidade Federal de Minas Gerais (Brazil)
Samuel Faria - Universidade Federal de Minas Gerais (Brazil)
Mauricio Castro - Pontificia Universidad Catolica de Chile (Chile)
Abstract: Epidemic modeling consists of the specification of an underlying structure that could rely entirely on epidemiological reasoning, be data-driven or a combination of them. In any case, it is based on the identification of characteristics that are shared by many regions. Some of these features present similarities across observational units. Hierarchical modeling is particularly useful in these settings as it allows the explicit incorporation of these similarities, thus enabling borrowing information across regions. The resulting setup is suitable for the estimation of the epidemic evolution and prediction of future epidemic cases. A number of options are considered, including those taking spatial configuration into account. These ideas are illustrated in the analysis of the evolution of Covid19 in Brazil, integrated across its 27 states.