Multiple signal integration by decision tree induction to detect artifactsin the neonatal intensive care unit

The high incidence of false alarms in the intensive care unit (ICU) necessi tates the development of improved alarming techniques. This study aimed to detect artifact patterns across multiple physiologic data signals from a ne onatal ICU using decision tree induction. Approximately 200 h of bedside da ta were analyzed. Artifacts in the data streams were visually located and a nnotated retrospectively by an experienced clinician. Derived values were c alculated for successively overlapping time intervals of raw values, and th en used as feature attributes for the induction of models trying to classif y 'artifact' versus 'not artifact' cases. The results are very promising, i ndicating that integration of multiple signals by applying a classification system to sets of values derived from physiologic data streams may be a vi able approach to detecting artifacts in neonatal ICU data. (C) 2000 Elsevie r Science B.V. All rights reserved.