A system developed by staff at the School of Informatics in conjunction with the College of Medicine and Veterinary Medicine, aims to prevent false alarms and detect danger signs at an earlier stage in the monitoring of premature babies. It is vital that premature babies are closely monitored to ensure that they receive rapid medical attention when required. However, one of the main problems with the current systems used for monitoring them is the number of false alarms which can be caused for example, by the baby being handled, or moving itself and dislodging the monitoring probes.
The system takes data from all probes and looks for patterns that indicate something is actually wrong, using probabilistic modelling. For example if the data shows that the baby’s heart has stopped suddenly then it is likely that a monitoring wire has come off: it is impossible for a baby’s heart to drop to zero in an instant, it is always a gradual slowing down. The system can also look at data from different probes to create a more detailed picture.
Reducing the number of false alarms could not only save valuable time for medical staff but it could also ensure that when the alarm does go off it is taken more seriously. For more information please see:
This article was published on Dec 22, 2010