Specialized MLP Classifiers to Support the Isolation of Patients Suspected of Pulmonary Tuberculosis

Abstract

Tuberculosis is an infectious disease widely present in developing countries, which is largely motivated by the difficulty of a rapid and efficient diagnosis. In order to reduce the number of patients suspected of having TB unnecessarily isolated in hospitals, thus optimize the use of health resources, we propose a systematic procedure for developing a decision support system based on specialized MLP network committee. The system based on 3 MLP models, which response to input data clusters inferred by the k-means technique, exhibits a better classification performance than a single network in terms of the number of false positives, achieving a sensitivity of 83.3% and specificity of 94.3%.