Knowledge Discovery in Databases for Prediction of Lab Tests Necessity

An ICU database collected at Hospital da Luz, in Lisbon was used for Knowledge retrieval. The Knowledge Discovery in Databases process is applied to this dataset, to obtain classification models that will allow to identify the laboratory testing necessity. Data preprocessing and reduction, feature selection, and classification processes are used. For the classification step, soft computing tools such as fuzzy systems and neural networks are used, combined with SFS, AFS and GA feature selection tools. The feature selection tools have been altered to work with distributed computing, in order to decrease the execution time.