Iron value classification in patients undergoing continuous ambulatory peritoneal dialysis using data mining

Abstract

In this article, Data Mining classification techniques are employed, in order to classify as normal or not-normal the iron values from a patients’ blood analysis. The dataset used is relative to patients that were subjected to Continuous Ambulatory Peritoneal Dialysis (CAPD) treatment. Weka software was used for testing several classification algorithms into such data set. The main purpose is finding the best suitable classification algorithm, with a pleasing performance in classifying the instances of the data, whereas preserving low rate of false positives. The IBk algorithm achieved the best performance, being able to correctly classify 97.39% of the instances.