An architecture for multi-dimensional temporal abstraction and its application to support neonatal intensive care

Temporal abstraction (TA) provides the means to instil domain knowledge into data analysis processes and allows transformation of low level numeric data to high level qualitative narratives. TA mechanisms have been primarily applied to uni-dimensional data sources equating to single patients in the clinical context. This paper presents a framework for multi-dimensional TA (MDTA) enabling analysis of data emanating from numerous patients to detect multiple conditions within the environment of neonatal intensive care. Patient agents which perform temporal reasoning upon patient data streams are based on the event calculus and an active ontology provides a central knowledge core where rules are stored and agent responses accumulated, thus permitting a level of multi-dimensionality within data abstraction processes. Facilitation of TA across a ward of patients offers the potential for early detection of debilitating conditions such as Sepsis, Pneumothorax and Periventricular Leukomalacia (PVL), which have been shown to exhibit advance indicators in physiological data. Preliminary prototyping for patient agents has begun with promising results and a schema for the active rule repository outlined.