Nova arquitetura utilizando regras de combinação fuzzy para métodos de aglomeração espacial aplicada à epidemiologia

• Ano de defesa: 2019
• Autor(a) principal: Sá, Laisa Ribeiro de
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal da Paraíba; Brasil; Ciências Exatas e da Saúde; Programa de Pós-Graduação em Modelos de Decisão e Saúde; UFPB
• Palavras-chave em Português: Métodos de aglomeração espacial; Epidemiologia; Conjuntos fuzzy; Regra de votação fuzzy; Combinação de classificadores; Spatial clustering methods; Epidemiology; Fuzzy sets; Fuzzy voting rule; Combination of classifiers; CNPQ::CIENCIAS DA SAUDE::SAUDE COLETIVA
• Link de acesso: https://repositorio.ufpb.br/jspui/handle/123456789/19432

In the health area, especially in epidemiology, methods that can georeference a specific event are needed, given that the information obtained from them is essential for the understanding of the phenomenon and for the decision-making process. The combination of spatial clustering methods, with the inclusion of the uncertainties presented in the health-disease process, produces better results when compared to their individual use, in order to produce more reliable information and more precise decision-making. The objective of this study was to develop a new architecture using fuzzy combination rules for spatial clustering methods applied to epidemiology. This is an ecological, retrospective, quantitative approach. The new architecture was tested in three different case studies using epidemiological data from Dengue cases reported in the state of Paraíba, Brazil, from the years 2011 and 2017. The case studies demonstrated the usability of the proposed architecture through the grouping of methods using different rules of combination, in addition to the comparison with previous architecture in order to point improvements in the results. The first case study carried out the combination of methods through weighted majority voting, resulting in a map of the state where municipalities were labeled as a priority, transitional, and non-priority, these classes were divided by the output value for each locality. The second case study had functions of aggregation as a combinator, classifying the municipalities as significant and not significant for the occurrence of Dengue. In the last case study, voting rules based on the cardinality of fuzzy sets were used, making it possible to classify municipalities as significant or nonsignificant, and to construct a map that divided municipalities into classes of degrees of pertinence. It should be noted that in all cases there was information fuzzification. Therefore, the result obtained with the application of the architecture using fuzzy combination rules for spatial clustering methods allowed the visualization of the spatial distribution of Dengue in all the municipalities of Paraíba, directing the managers to decisions that respect the particularities of each place, in addition to being a low-cost method.