• Entrar
    Ver item 
    •   Página inicial
    • Teses & Dissertações
    • Teses & Dissertações
    • Ver item
    •   Página inicial
    • Teses & Dissertações
    • Teses & Dissertações
    • Ver item
    JavaScript is disabled for your browser. Some features of this site may not work without it.

    Aplicação de rede neural artificial para o reconhecimento do diabetes mellitus gestacional com marcadores não-glicêmicos

    Thumbnail
    Visualizar/Abrir
    R - D - ANA PAULA FILUS BANDEIRA.pdf (1.485Mb)
    Data
    2015
    Autor
    Bandeira, Ana Paula Filus
    Metadata
    Mostrar registro completo
    Resumo
    Resumo: O Diabetes mellitus gestacional (DMG) afeta cerca de 7% das gestações e tem impacto importante para a gestante, o feto e o neonato. O diagnóstico do DMG é realizado com base na determinação da glicemia em jejum e após sobrecarga. A identificação de padrões com base em características antropométricas e laboratoriais pode ser de interesse no processo de triagem do DMG. Nesse trabalho, o toolbox, open source, "WEKA" foi empregado, utilizando os algoritmos simple k-means e simple logistic, para a classificação de variáveis não-glicêmicas. A rede neural artificial, multilayer perceptron (MLP), foi aplicada na busca da identificação automática do DMG. A base do estudo foi uma amostra de 997 gestantes (699 = gestantes saudáveis (controle); 298 = gestantes com diabetes gestacional (DMG)) utilizando critérios glicêmicos estabelecidos pela American Diabetes Association. Os marcadores discriminantes selecionados com auxílio dos algoritmos k-means e simple logistic foram idade, peso, índice de massa corporal, pressão arterial sistólica, pressão arterial diastólica, ácido úrico, triglicérides, colesterol não HDL e Log(TG/HDL-C). Estas variáveis foram aplicadas na rede neural MLP para a classificar gestantes com DMG. A MLP devidamente treinada e testada permitiu 88% de predições corretas, estabelecendo uma sensibilidade de 92,1%, especificidade de 83,8% e acurácia de 87,5%. Em síntese, a metodologia desenvolvida neste trabalho é de baixo custo e uma alternativa de "segunda opinião" para a triagem do DMG. Palavras-chave: Diabetes mellitus gestacional; sistemas de apoio à decisão clínica; WEKA, k-means, simple-logistic, redes neurais artificiais.
     
    Abstract: The gestational diabetes mellitus (GDM) affects about 7% of pregnancies and has an important impact on the mother, the fetus and the newborn. The diagnosis of GDM is based on the determination of fasting and post-load glycemia. The identification of patterns based on anthropometric and laboratory parameters can be of interest in the screening process of DMG. In this work, the toolbox, open source, "WEKA" was employed, using the simple k-means and simple logistic algorithms, for non-glycemic variables rating. The artificial neural network, multilayer perceptron (MLP), was applied in the search for automatic identification of the DMG. The basis of the study was a sample of 997 patients (699 = healthy pregnant women (control); 298 = pregnant women with gestational diabetes (GDM)) using glycemic criteria established by the American Diabetes Association. The discriminant markers selected with the help of k-means and simple logistic algorithms were age, weight, body mass index, systolic blood pressure, diastolic blood pressure, uric acid, triglycerides, non-HDL cholesterol and Log(TG/HDL-C). These variables were applied in MLP to classify pregnant women with GDM. The MLP network properly trained and tested allowed 88% of correct predictions, establishing a sensitivity of 92.1%, specificity of 83.8% and accuracy of 87.5%. In summary, the methodology developed in this work is low cost and an appropriate alternative to "second opinion", allowing GDM screening. Keywords: gestational diabetes; decision support systems; WEKA; k-means; simple logistic; artificial neural networks.
     
    URI
    https://hdl.handle.net/1884/38910
    Collections
    • Teses & Dissertações [10807]

    DSpace software copyright © 2002-2022  LYRASIS
    Entre em contato | Deixe sua opinião
    Theme by 
    Atmire NV
     

     

    Navegar

    Todo o repositórioComunidades e ColeçõesPor data do documentoAutoresTítulosAssuntosTipoEsta coleçãoPor data do documentoAutoresTítulosAssuntosTipo

    Minha conta

    EntrarCadastro

    Estatística

    Ver as estatísticas de uso

    DSpace software copyright © 2002-2022  LYRASIS
    Entre em contato | Deixe sua opinião
    Theme by 
    Atmire NV