A clinical grid infrastructure supporting adverse hypotensive event prediction

Abstract

The condition of hypotension - where a person's arterial blood pressure drops to an abnormally low level - is a common and potentially fatal occurrence in patients under intensive care. As medical interventions to treat such events are typically reactive and often aggressive, there would be great benefit in having a prediction system that can warn health-care professionals of an impending event and thereby allow them to provide non-invasive, preventative treatments. This paper describes the progress of the EU FP7 funded Avert-IT project, which is developing just such a system using Bayesian neural network learning technology based upon an integrated, real-time data grid infrastructure, which draws together heterogeneous data-sets from six clinical centres across Europe.