Título: Sistema para Identificação de Padrões do Sono com o auxílio de Redes Neurais Dinâmicas
Autores: Rodrigues, Marco Aurélio Benedetti; Azevedo, Fernando Mendes de; Marino Neto, José
Resumo: The present work aims to identify sleep patterns identifications through the application of Dynamic Artificial Neural Networks (DANN). This one is very important in the identification and classification of the sleep stages. The input patterns of the network was obtained of an EEG registration of a patient during the sleep. The group of training patterns for DANN was constituted of 5 spindles of the sleep and 5 K Complex and 5 slow waves rhythms, characteristic of the two sleep phase. The results obtained in the sleep patterns identification with DANN were compared with the results obtained by a Static Artificial Neural Network (SANN), with the purpose of showing th e superiority of DANN. These training patterns were previously classified by a specialist, which agreed with approximately 90% of the identified patterns.
Artigo em pdf: 4cbrn_077.pdf
Arquivo BibTex: 4cbrn_077.bib