Big Data Analytical Technologies and Decision Support in Critical Care

Abstract

Intensive care units use sophisticated patient monitoring systems that produce massive amounts of physiological streaming data. Critical care physicians manage several hundred of time-critical health-related variables a day in the course of their work, using data from multiple disjoint systems, running in silos. Patient monitoring devices produce hundreds of alarms per day for each patient, with approximately 90 % being insignificant. These have introduced a significant data and decision overload problem. We present state- of-the-art big data analytical approaches that address these problems. We describe how the combination of “at rest” data mining analytics and streaming analytics is transforming critical care by enabling applications that improve clinicians’ situation awareness at the bedside. We describe novel system architectures together with real-world deployments of big data analytical technologies in critical care environments that are helping physicians to be more proactive in delivering timely care.