Título: Árvore de Classificação e Redes Neurais Artificiais: Uma Aplicação à Predição de Tuberculose Pulmonar
Autores: Santos, Alcione Miranda; Pereira, Basilio de Bragança
Resumo: Smear negative pulmonary tuberculosis (SNPT) accounts of 30% of Pulmonary Tuberculosis (PT) cases reported yearly. Rapid and accurate diagnosis of SNPT could provide lower morbidity and mortality, and case detection at a less contagious status. The main objective this work is to evaluate a prediction model for diagnosing SNPT, useful for outpatients attended in settings with limited resources. One hundred thirty six patients from Health Care Units were included. They were referred to our Teaching Hospital, in Rio de Janeiro, Brazil, from March, 2001 to September, 2002, with clinical-radiological suspicion of SNPT. Only symptoms and physical signs were used for constructing the neural network (NN) and the classification tree. The NN classified correctly 73% ofpatients from the test sample, while the classification tree classified only 40% of those patients. The NN model suggests that mathematical modeling, for classifying SNPT cases, could be a useful tool for optimizing utilization of more expensive tests, and to avoid costs of unnecessary anti-PT treatment.
Código DOI: 10.21528/CBRN2003-091
Artigo em PDF: 6CBRN_091.PDF
Arquivo BibTex: 6CBRN_091.bib