Tratamento Eficiente da Incerteza em Sistemas de Apoio à Decisão

Cecília Dias FloresCharles Leandro HoherMarcelo LadeiraRosa Maria Vicari

This paper presents the SEAMED framework. Real domains into environment with uncertainty can be model with this framework, as follows. The theory of probabilities is used to model and to treat the inherent uncertainty of the domain. The domain is modeled with a set of random variables. A directed link from a random variable to another one represents the direct relationship of likelihood, conditioning, relevance or causality among these random variables. The probability distribution function of a random variable, giving its parents, represents the strengths of the relationship among the variable and its parents. The SEAMED presents a graphical interface designed to ease the construction of decision-making support applications into some medical fields. In order to analyze a specific case, the user of the application should enter the total evidence (information) available. Then, the application propagates this evidence through the other random variables and updates theirs probabilities giving the evidence available into the system. Generally, the medical reasoning is of the diagnostic kind, that is, the physician observes the effects caused by diseases in a patient and the results of exams in order to find out the most provable cause pathologies. The SEAMED evidence entrance follows the medical reasoning metaphor, that is, it is based on the anamnesis (patient's clinical history, patient's major complain and realization of physic exams), the carry out of complementary exams, and the performance of differential diagnoses.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: