Redes bayesianas dinâmicas com definição de limiar aplicadas ao estudo de caso detecção de extrassístole ventricular

• Ano de defesa: 2010
• Autor(a) principal: Oliveira, Lorena Sophia Campos de
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Espírito Santo; BR; Doutorado em Engenharia Elétrica; Centro Tecnológico; UFES; Programa de Pós-Graduação em Engenharia Elétrica
• Palavras-chave em Português: Redes bayesianas; Complexos ventriculares prematuros; Técnicas de apoio para a decisão; Inteligência artificial; Incerteza; Teoria bayesiana de decisão estatística; Engenharia Elétrica; 621.3
• Link de acesso: http://repositorio.ufes.br/handle/10/9695

This work proposes a dynamic bayesian approach with threshold setting to develop a system to support a cardiologist in making a decision, in terms of classifying a heart beat. Dynamic Bayesian Networks (DBN) and static Bayesian Networks (BN) are adopted for performing such classification, since they are very suitable to deal with the uncertainties involved in the cardiologist‟s reasoning, thanks to their probabilistic and logic model. Different BN topologies are implemented and tested, aiming at finding the one more suitable to the problem under study. Specifically speaking, it is considered the detection of premature beats (PVC), which are one kind of arrhythmia related to the premature contraction of the ventricles. The results obtained with the use of Dynamic Bayesian Network with threshold setting for detection of premature beats reached 99.53%, 100%, 100% and 99.97% for the values of Sensitivity, Specificity, Positive Predictive Value and Negative Predictive Value respectively, and confidence interval of ± 0.4% and ± 0% for sensitivity and positive predictive value respectively.