A Framework for Processing Cumulative Frequency Queries over Medical Data Streams

Medical data streams processing becomes increasingly important since it extracts critical information from a continuous flow of patient data. Various types of problems have been studied on medical data streams, such as classification, clustering, anomaly detection, etc.; however, efficient evaluation of cumulative frequency queries has not been well studied. The cumulative frequency of patients' status can play an instrumental role in monitoring the patients' health conditions. Up to now, efficiently processing cumulative frequency queries on medical data streams is still a challenging task due to the large size of the incoming data. Therefore, in this paper, we propose a novel framework for processing the cumulative frequency queries over medical data streams to support the online medical decision. The proposed framework includes two components: data summarisation and dynamic maintenance. For data summarisation, we propose a hybrid approach that combines two data structures and exploits a classification algorithm to select the more efficient data structure for computing the cumulative frequency. For dynamic maintenance, we propose an incremental maintenance approach for updating the cumulative frequencies when new data arrive. The experimental results on a real dataset demonstrate the efficiency of the proposed approach.