A framework to model and translate clinical rules to support complex real-time analysis of physiological and clinical data

We present a framework to model and translate clinical rules to support complex real-time analysis of both synchronous physiological data and asynchronous clinical data. The framework is demonstrated through a case study in a neonatal intensive care context showing how a clinical rule for detecting an apnoeic event is modeled across multiple physiological data streams in the Artemis environment, which employs IBM's InfoSphere Streams middleware to support real-time stream processing. Initial clinical hypotheses for apnoea detection are modeled using UML activity diagrams which are subsequently translated into Stream's SPADE code to be deployed in Artemis to deliver real-time decision support. Our aim is to provide a Clinical Decision Support System capable of identifying and detecting patterns in physiological data streams indicative of the onset of clinically significant conditions that that may adversely affect health outcomes. Benefits associated with our approach include: 1) reduced time and effort on the clinician's part to assess health data from multiple sources; 2) the ability to allow clinicians to control the rules-engine of Artemis to enhance clinical care within their unique environments; 3) the ability to apply clinical alerts to both synchronous and asynchronous data; and 4) the ability to continuously process data in real-time.