Reconhecimento de padrão na biodisponibilidade do ferro utilizando o Algoritmo Fuzzy C-Means

Abstract

This dissertation presents a method for pattern recognition on the bioavailability of iron, through interaction with substances that help the absorption such as vitamin C and vitamin A and inhibitors as calcium, phytate, oxalate, tannin and caffeine. The database was obtained through dietary, lunch and dinner, in children 2-5 years in the Municipal Nursery of Paraty - Rio de Janeiro, between 2007 and 2008. The Principal Component Analysis (PCA) was applied in the selection of nutrients and used the Fuzzy C-Means Algorithm (FCM) to create the groups classified according to the bioavailability of iron. A sensitivity analysis was developed in an attempt to find neighboring amounts of calcium being consumed at meals. The PCA showed that at lunch the nutrients that best explained the variability of the model were iron, vitamin C, phytate and oxalate, while at dinner the calcium was effective in determining the variability of the model due to high consumption of dairy products. For lunch, the application of FCM in the interaction of nutrients, it was noted that the intake of vitamin C was decisive in the classification of groups. At dinner, the classification of groups was determined by the amount of iron in the interaction with calcium. In the sensitivity analysis performed for lunch and dinner, two iterations of the algorithm determined the total interference of calcium on iron bioavailability.