Monitoring abnormal patterns with complex semantics over ICU data streams

Abstract

Monitoring abnormal patterns in data streams is an important research area for many applications. In this paper we present a new approach MAPS (Monitoring Abnormal Patterns over data Streams) to model and identify the abnormal patterns over the massive data streams. Compared with other data streams, ICU streaming data have their own features: pseudo-periodicity and polymorphism. MAPS first extracts patterns from the online arriving data streams and then normalizes them according to their pseudo-periodic semantics. Abnormal patterns will be detected if they are satisfied the predicates defined in the clinician-specifying normal patterns. At last, a real application demonstrates that MAPS is efficient and effective in several important aspects. © Springer-Verlag Berlin Heidelberg 2006.