Rede neural artificial: um modelo de apoio à decisão em segurança alimentar para municípios do interior da Paraíba

• Ano de defesa: 2013
• Autor(a) principal: Silva, Cleyton Cézar Souto
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal da Paraí­ba; BR; Ciências Exatas e da Saúde; Programa de Pós-Graduação em Modelos de Decisão e Saúde; UFPB
• Palavras-chave em Português: Segurança alimentar; Rede neural artificial; Modelos de decisão; Food security; Artificial neural network; Decision models; CIENCIAS DA SAUDE::SAUDE COLETIVA
• Link de acesso: https://repositorio.ufpb.br/jspui/handle/tede/6545

Food insecurity exists when the availability of nutritionally adequate and safe foods or the ability to acquire them in a socially acceptable is limited or uncertain. Therefore, the prevalence of food insecurity is of great importance for the assessment of living conditions and, consequently, for the planning of public policies to fight hunger. So it is timely and relevant indicators that can create check from the right of access to food until the concrete conditions of such access and its ultimate effects on health and nutrition of individuals and collective activities. Thus, it is intended by an artificial neural network model for decision support in food security and assist in identifying the severity of this situation in the populations of the cities of São José dos Ramos and the Nova Floresta in the interior of Paraíba. It is a population-based cross-sectional study with a sample of 618 households in the two counties, and 287 in São José dos Ramos and 331 in Nova Floresta. For this study the measurement of the rate of food insecurity and its levels were estimated using the methodology of the Brazilian Food Insecurity Scale for model creation and decision support data were grouped as mild-moderate food insecurity and severe food insecurity. We selected 10 quantitative variables on socioeconomic and demographic: number of rooms in the house; numbers used rooms in the house to sleep, total household members; schooling in years of the household head, number of children and adolescents attending school , number of children, number of adolescents, number of adults, number of elderly, relationship between workers and unemployed and with the help of MATLAB software generates a model of Artificial Neural Network feedforward with one input layer, one hidden layer with 22 neurons , and 1 output layer with 2 neurons with backpropagation learning. Based on this, the generated models achieved the following results: 81% correct in deciding on food security and food insecurity x 80.2% of food insecurity in the decision mild-moderate to severe insecurity x São José dos Ramos; Nova Floresta got 80. 7% correct in deciding on food security and food insecurity x 80.4% food insecurity in the decision to take x-moderate severe insecurity. Therefore, this model constitutes an important tool to define the trends priority intervention in municipalities, in order to permit the identification of this disease at the local level and can support the process of decision making and planning of public policies and actions aimed at promoting food security.