Adaptive API for Real-Time Streaming Analytics as a Service.

Publication Type:
Conference Proceeding
Conference proceedings : ... Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual Conference, 2019, 2019, pp. 3472-3477
Issue Date:
Full metadata record
A significant amount of physiological data is generated from bedside monitors and sensors in neonatal intensive units (NICU) every second, however facilitating the ingestion of such data into multiple analytical processes in a real time streaming architecture remains a central challenge for systems that seek effective scaling of real-time data streams. In this paper we demonstrate an adaptive streaming application program interface (API) that provides real time streams of data for consumption by multiple analytics services enabling real-time exploration and knowledge discovery from live data streams. We have designed, developed and evaluated an adaptive API with multiple ingestion of data streamed out of bedside monitors that is passed to a middleware for standardization and structuring and finally distributed as a service for multiple analytical services to consume and perform further processing. This approach allows, (a) multiple applications to process the same data streams using multiple algorithms, (b) easy scalability to manage diverse data streams, (c) processing of analytics for each patient monitored at the NICU, (d) ability to integrate analytics that seek to evaluate multiple patients at the same point in time, and (e) a robust automated process with no manual interruptions that effectively adapts to changing data volumes when bedside monitors increases or the amount of data emitted by a monitor changes. The proposed architecture has been instantiated within the Artemis Platform which provides a framework for real-time high speed physiological data collection from multiple and diverse bed side monitors and sensors in NICUs from multiple hospitals. Results indicate this is a robust approach that can scale effectively as data volumes increase or data sources change.
Please use this identifier to cite or link to this item: