Design of an Intelligent Intensive Care Unit Alarm System

The Intensive Care Unit (ICU) accommodates the most severely ill patients in a hospital. These patients are monitored by several medical devices, e.g. ventilators and gas monitors. All of these devices function autonomously giving individual alarms. As a result alarms are often triggered unnecessarily, e.g. in the event of a probe failing, even though another device may measure that the patients condition is unchanged. This article investigates the possibility of processing alarms intelligently through the development of a system, that is capable of evaluating alarms based on data from several medical monitoring devices. This involves detecting equipment failure alarms, trend alarms, and threshold alarms. Furthermore, it is desirable for the hospital staff to be able to monitor the patient from any location in the hospital. The emphasis of this article is on developing a monitoring system, providing a network of distributed data collection units, and the ability to monitor measurements and alarms from any given location within the network. The system is comprised of three parts: A bedside unit for each patient, multiple independent monitoring units, and a central data repository. The bedside unit collects data from medical devices monitoring the patient. This data is distributed via TCP/IP to all connected monitoring units and the central data repository. The monitoring unit has three central components: The discovery, the graphical user interface (GUI), and the alarm component. When started, available bedside units are discovered through multicasting. When connected to a bedside unit the monitoring unit displays the collected data and alarms. The central data repository stores measured data from the medical devices connected to the bedside units. The resulting design has proven successful in evaluating alarms based on data from several medical monitoring devices. Test cases show that the design can differentiate the three different alarm types. The design has also proven successful in distributing the measured data from one bedside unit to multiple monitoring units. This article demonstrates the possiblity of generating intelligent alarms in an ICU by implementing a distributed system, with a bedside unit for each patient and a selfcontained monitoring unit. The flexibility of the system can be expanded considerably by implementing XML as the data exchange format between nodes in the system. Furthermore, more advanced alarm types, e.g. decision support systems, can be implemented.