Ensemble methods for meningitis aetiology diagnosis

In this work, we explore data-driven techniques for the fast and early diagnosis concerning etiological origin of meningitis, more specifically with regard to differentiating between viral bacterial meningitis. We study how machine learning can be used predict meningitis aetiology once a patient has been diagnosed disease. have dataset 26,228 patients described by 19 attributes, mainly about patient's observable symptoms results cerebrospinal fluid analysis. Using dataset, explored several sampling, feature selection classification models based both on ensemble methods simple (mainly, decision trees). Experiments 27 (19 them involving methods) conducted paper. Our main finding is that combination trees leads best classifiers. The performance indicator values (precision, recall f-measure 89% an AUC value 95%) achieved synergy bagging NBTrees. Nonetheless, our also suggest certain tree clearly improves in comparison those obtained only corresponding tree.