An integrated data management approach to manage health care data

Intensive Care Unit data management systems suffer from three problems: data and meta-data are spread out in different systems, there is a high rate of false positives due to default thresholds, and data mining predictions are not available in a timely manner. This proof-of-concept demonstration, based on the Intensive Care Unit environment of the University of Utah Health Sciences Center, presents a system that: i) integrates in one place historical data, events, rules, and data mining models; ii) is highly customizable letting users create or change rules; and iii) identifies possible future risks by performing data mining in soft-real-time. Using simulated inputs, we show the complete system working, including writing and editing rules, triggering simple alerts, prediction of cardiac arrests, and visual explanation of predictions