Using Temporal Constraints to Integrate Signal Analysis and Domain Knowledge in Medical Event Detection
The events which occur in an Intensive Care Unit (ICU) are many and varied. Very often, events which are important to an understanding of what has happened to the patient are not recorded in the electronic patient record. This paper describes an approach to the automatic detection of such unrecorded ’target’ events which brings together signal analysis to generate temporal patterns, and temporal constraint networks to integrate these patterns with other associated events which are manually or automatically recorded. This approach has been tested on real data recorded in a Neonatal ICU with positive results.
KeywordsDomain Knowledge Neonatal Intensive Care Unit Discrete Event Temporal Constraint Target Event
Unable to display preview. Download preview PDF.
- 2.Fried, R., Gather, U.: Online Pattern Recognition in Intensive Care Medicine. In: Proceedings of AMIA Annual Symposium, pp. 184–188 (2001)Google Scholar
- 3.Chambrin, M.C., Charbonnier, S., Sharshar, S., Becq, G., Badji, L.: Automatic Charac-terization of Events on SpO2 signal: Comparison of Two Methods. In: Proceedings of the 26th Annual International Conference of the IEEE EMBS, pp. 3474–3477 (2004)Google Scholar
- 4.Portet, F., Gao, F., Hunter, J., Sripada, Y.: Evaluation of On-line Bradycardia Boundary Detectors from Neonatal Clinical Data. In: 29th Annual International Conference of the IEEE EMBS, pp. 3288–3291 (2007)Google Scholar
- 5.Quinn, J.A., Williams, C.K.I., McIntosh, N.: Fact0orial Switching Kalman Filters Ap-plied to Condition Monitoring in Neonatal Intensive Care. In: IEEE Transactions on Pattern Analysis and Machine Intelligence (to appear)Google Scholar
- 10.Hunter, J., Ewing, G., Freer, Y., Logie, R., McCue, P., McIntosh, N.: Neonate: Decision Support in the Neonatal Intensive Care Unit - A Preliminary Report. In: 9th European Con-ference on Artificial Intelligence in Medicine, pp. 41–45 (2003)Google Scholar