Aperfeiçoamento e Ajuste Paramétrico de Modelo Baseado em Agentes para Simulação da Transmissão da Dengue

• Ano de defesa: 2022
• Autor(a) principal: Cunha, Eduardo Argou Aires
• Orientador(a): Rizzi, Rogério Luis Rizzi
• Banca de defesa: Galante, Guilherme , Rizzi, Claudia Brandelero, Alves, Fernando Lima
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Estadual do Oeste do Paraná; Cascavel
• Programa de Pós-Graduação: Programa de Pós-Graduação em Ciência da Computação
• Departamento: Centro de Ciências Exatas e Tecnológicas
• País: Brasil
• Palavras-chave em Português: Transmissão da dengue; Modelo baseado em agentes; Algoritmo genético; Métodos de Monte Carlo; Ajuste paramétrico
• Palavras-chave em Inglês: Dengue transmission; Agent-based model; Genetic algorithm; Monte Carlo methods; Parameter adjustment
• Área do conhecimento CNPq: MÉTODOS EM COMPUTAÇÃO APLICADA
• Link de acesso: https://tede.unioeste.br/handle/tede/6401

This work presents an agent-based system developed in order to simulate scenarios of epidemics to study dengue propagation. Such system was conceived by means of the modelling and implementation of movement, contact, transition and control operators that characterize an agent’s evolution over time in a bidimensional environment which houses human and mosquito population, specifically Aedes aegypti. The main goal of this work is the improvement of the model and obtaining a satisfactory parameter configuration for the simulator, such that the response variables of its output are close to real observed data. The referred scenario is a dengue epidemic occured in the municipality of Cascavel-PR in the years of 2015 and 2016. Two distinct scenarios were considered for experimental evaluation, one being the global dynamics envolving the entire environment, the other consisting of a subset of the global scenario limited to a smaller region only. This work presents strategies for spatial representation and evaluation of the distribution of simulated cases which may be useful as new response variables to be considered when applying ajustment techniques. Also three alternatives were implemented for the movement routine of the female mosquito, including several comparative studies for them. A Genetic Algorithm (GA) was implemented for the parameters adjustment and this approach was able to adjust 3 parameters to values which results are near the real data for both scenarios, albeit it is still complex to use the obtained values for any kind of interpretation. The application of the GA yielded good results with different rules of reproduction using arithmetic and geometric means, also with promising performance in the inclusion of a fourth parameter, all of which indicate that more sophisticated genetic and evolutive approaches might be able to produce even better results.