Artificial Neural Networks and Bayesian Networks as supportting tools for diagnosis of asymptomatic malaria

Abstract

In the preset study, Artificial Neural Network (ANN) and Bayesian Network (BN) techniques are evaluated as supporting tools for the diagnosis of asymptomatic malaria infection. These techniques are compared with two classical laboratorial tests for diagnosis of malaria: the light microscopy and the Nested PCR. To do this, the tests were run in a group of 380 individuals from the Brazilian Amazon. The results indicate that both innovative techniques are able to identify asymptomatically infected individuals with better accuracy than the microscopy test and are potentially useful for helping the diagnosis of asymptomatic malaria.