Analyzing dengue epidemic dynamics using Physics-Informed Neural Networks

Resumo

This thesis presents a novel approach for modeling dengue epidemics through an adaptation of Physics-Informed Neural Networks (PINNs) designed to model dengue dynamics with time-varying parameters. By incorporating time-varying parameters within a deterministic nonlinear differential equation framework, the approach is particularly well-suited for modeling the complex and evolving nature of dengue transmission, which is marked by serotype variations, changes in population susceptibility, and temporal shifts in transmission dynamics. By applying the model to dengue data from Rio de Janeiro, Goiania, Fortaleza, and Joinville, we examine the role of these parameters in different epidemic scenarios, revealing unique patterns tied to local epidemiological and climatic factors. This city-specific analysis highlights the framework's capacity to capture the interplay of local epidemiology. Furthermore, this study paves the way for future work that integrates climate data into the model, allowing us to simulate disease dynamics under various climate change scenarios.This thesis presents a novel approach for modeling dengue epidemics through an adaptation of Physics-Informed Neural Networks (PINNs) designed to model dengue dynamics with time-varying parameters. By incorporating time-varying parameters within a deterministic nonlinear differential equation framework, the approach is particularly well-suited for modeling the complex and evolving nature of dengue transmission, which is marked by serotype variations, changes in population susceptibility, and temporal shifts in transmission dynamics. By applying the model to dengue data from Rio de Janeiro, Goiania, Fortaleza, and Joinville, we examine the role of these parameters in different epidemic scenarios, revealing unique patterns tied to local epidemiological and climatic factors. This city-specific analysis highlights the framework's capacity to capture the interplay of local epidemiology. Furthermore, this study paves the way for future work that integrates climate data into the model, allowing us to simulate disease dynamics under various climate change scenarios.

Ficha do documento

Tipo

Tese

Ano

2024

Instituição

Fundação Getulio Vargas

Fonte

Repositório da FGV

Idioma

Inglês

Acesso

Não informado

Identificador

oai:repositorio.fgv.br:10438/36468

Temas

Meio Ambiente

Palavras-chave

Physics-Informed Neural Networks (PINNs)Time-varying parametersDengue transmissionMatemáticaEquações diferenciaisRedes neurais (Computação)DengueFatores climáticos