Diagnosis of Headaches Types Using Artificial Neural Networks and Bayesian Networks
Keywords:
Machine learning; artificial intelligence; headache diagnosisAbstract
The aim of this paper is to classify several types of headaches from patients, using different data analysis methods, with the application of two classifying algorithms, comparatively: Bayesian Networks (BN) and Artificial Neural Networks (ANN). The data was collected through a data survey process applied to 2,177 patients with headache diagnosis in Neurological Clinic of municipality of Joinville/SC, Brazil, from January 2010 to November 2014. It was constructed eight different Test Models, varying: the attributes codification (codification A and B); the network output (one and five) and the classifier algorithms (BN and ANN). It is presented the accuracy results obtained from the eight Test Models, using the software WEKA® (Waikato Environment for Knowledge Analysis). The results presented a good accuracy in all tests realized and they suggest that BNs give better accuracy when comparing to ANNs for this problem.