The University of Edinburgh -
Division of Informatics
Forrest Hill & 80 South Bridge

MSc Thesis #94121

Title:Detection of Data Trends in Computerised Cot Monitoring
Date: 1994
Abstract:In the Neonatal Unit of the Simpson Memorial Maternity Pavilion (SMMP), premature or sick newborn infants have various physiological parameters continuously monitored by probes placed upon the infant, to enable the neonatologists caring for these infants to spot impending problems. These parameters include heart rate, respirator rate, blood pressure, temperatures, and blood carbon dioxide and oxygen levels. This information is collected, stored, and displayed in graphical form by a computerised cot monitoring system. At present, simple threshold alarms operate on individual parameters to alert staff when a parameter value rises above/falls below its threshold. However, in practice a trend within the data for a single parameter, or specific combinations of such trends in various parameters, may be suggestive of clinical abnormalities although none of these data trends actually exceed their thresholds. A program which could identify such combinations of trends would thus be able to alert staff to a potential problem earlier that the existing system. We have developed a trend detection program which analyses the measured data for both blood carbon dioxide and blood oxygen levels and issues alerts whenever clinically significant trends are detected simultaneously over the same time window within the data for these two separate parameters. The system has been continuously evaluated throughout its evaluation by a domain expert at the SMMP and designed according to this expert's requirements, ascertained over several knowledge elicitation sessions.

[Search These Pages] [DAI Home Page] [Comment]