| Abstract | Keeping the oxygenation status of newborn infants within physiologic limits is a crucial task in intensive
care. For this purpose several vital parameters are supervised routinely by monitors, such as
electrocardiograph, transcutaneous partial oxygen pressure monitor and pulse oximeter. Each monitor
issues an alarm signal whenever an upper or lower limit of the parameter(s) measured is exceeded.
However, in practice it turns out, that a considerable amount of false alarms is generated by artefacts,
which are attributed mostly to movements of the infants. Eliminating these false alarms would be of benefit
to the staff as well as the patients of the intensive care unit. Accordingly, an automated system based on
Fuzzy Logic was developed, which is capable of distinguishing between critical situations and artefacts.
The system is based on a Transputer IMS T425 in a PC, which collects the data from the monitors, plots it
on a colour screen, saves it to hard disk and analyses it by Fuzzy Logic. Fuzzy algorithms were
developed to generate more reliable alarms. All vital parameters of eight infants, who either moved often
and/or frequently produced real alarm situations, were recorded. Synchronously the infants' movements
and care procedures were video taped. The data and video were analysed off line with the help of an
experienced neonatologist. His judgement was compared to the analysis of the Fuzzy Logic system. The
results show that it is possible to improve the reliability of the monitored data with the aid of an evaluation
strategy based on Fuzzy Logic and hence distinguish between real alarm situations and movement
artefacts to the extent that an application in an intensive care unit under routine conditions becomes
conceivable. |