Real-time data warehouse loading methodology and architecture: a healthcare use case

Abstract

In the healthcare context, existing systems suffer from the lack of supporting heterogeneity and dynamism. Consequently, resulting from sensors, streaming data brought another dimension to data mining research. This is due to the fact that, in data streams, only a time window is available. Contrary to the traditional data sources, data streams present new characteristics as being continuous, high-volume, open-ended and concept drift. To analyse event streams, data warehouse seems to be the answer to this problematic. However, classical data warehouse does not incorporate the specificity of event streams that are spatial, temporal, semantic and real-time. For these reasons, we focus inhere on presenting the conceptual modelling, the architecture and loading methodology of the real-time data warehouse by defining a new dimensionality and stereotype for classical data warehouse. To prove the efficiency of our real-time data warehouse, we adapt the model to a medical unit pregnancy care case study which show promising results.