A management strategy of monitor data in ICU based on data stream technology

Abstract

The data produced by such medical monitor devices as ventilators, cardiac monitors in the ICU, arrive in multiple, continuous, rapid, time-varying data streams. Such data can not be stored entirely because of the storage devices limitation and the interference data produced by doctors' or nurses' medical operations and patients' activities. The present ICU systems using a relational database to store and manipulate data only provides the "operator-dependent" way to collect these stream data at fixed time intervals set by doctors. In this way, some important monitor data may be ignored, but some interference data may be collected. So the quality of data management can not meet the needs of doctors' and nurses'. In this paper, after deeply analyzing the features of the stream monitor data in the ICU, we talk about using data stream technology to solve the problem. Before the stream data arriving, doctors will pre-establish some data collecting rules based on patients' illness status such as fixed data collecting time intervals, data alarming range and level, and so on. When the monitor data arriving, they will be analyzed and manipulated by the pre-defining query rules above pre-established. So it can provide a "patient-dependent" way to collect not only the monitor data affixed time intervals when patients in good condition, but also all the monitor data when they are in critically ill situation. Furthermore, in this paper we also provide a way to deal with the interference data produced by monitor devices. Data stream technology can supply a gap in the management of the stream monitor data which is based on a relational database, and improve the quality of the data management in the ICU.