Organ failure diagnosis by artificial neural networks

Abstract

In recent years, Clinical Data Mining has gained an increasing acceptance by the research community, due to its potential to find answers that could extend life or give comfort to ill persons. In particular, the use of tools such as Artificial Neural Networks, which have been mostly used in classification tasks. The present work reports the adoption of these techniques for the prediction of organ dysfunction of Intensive Care Unit patients. The novelty of this approach is due to the use intermediate outcomes, defined by the Out of Range Measurements of four bedside monitored variables, which obtained an overall accuracy of 70%.