Mobile Data Streams Research Articles

Prediction of Hospital Readmission from Longitudinal Mobile Data Streams.

Hospital readmissions impose an extreme burden on both health systems and patients. Timely management of the postoperative complications that result in readmissions is necessary to mitigate the effects of these events. However, accurately predicting readmissions is very challenging, and current approaches demonstrated a limited ability to forecast which patients are likely to be readmitted. Our research addresses the challenge of daily readmission risk prediction after the hospital discharge via leveraging the abilities of mobile data streams collected from patients devices in a probabilistic deep learning framework. Through extensive experiments on a real-world dataset that includes smartphone and Fitbit device data from 49 patients collected for 60 days after discharge, we demonstrate our framework’s ability to closely simulate the readmission risk trajectories for cancer patients.

Mining Social Distance from Mobile Phone Communication Networks

Nowadays, smartphones have been widely deployed. Scientists can obtain position and proximity data as well as call and SMS log data easily. Rich information are hidden behind these mobile data. For example, social interaction patterns of smartphone users. The discovery of those patterns can benefit many important applications such as (1) Relationship inference; (2) Community detection in counter terrorism. However, mobile data owns many unique characteristics, such as (1) The amount of data is massive; (2) Timeliness is very important; (3) Different types of data are highly coupled. Traditional statistical methods have been used to analyze mobile data, which can not be applied to huge amount of mobile data streams in real-world situations. In this paper, we take mobile data as an example, to analyze the social distance and fulfill popular tasks such as (1) Relationship inference; (2) Community detection. We propose a time-constrained homogeneous graph model, which takes a mobile call log as input and models the social distance problem by a time-based sliding window scheme. We further design an efficient modified EdgeRank algorithm to compute the social distances among users. We propose an algorithm for community detection. Experimental results on real data sets show that the proposed approach can infer social distance efficiently and achieves an accuracy of 80-85% automatically. We apply the results of social distance relationships to community detection and an improvement of accuracy of 5% can be achieved.

Design and Evaluation of an Autonomous Load Balancing System for Mobile Data Stream Processing Based On a Data Centric Publish Subscribe Approach

Several new applications of mobile computing environments, such as Intelligent Transportation Systems, Fleet Management and Logistics, and integrated Industrial Process Automation share the requirement of remote monitoring and high performance processing of huge data streams produced by large sets of mobile nodes. Two key requirements for the deployment and operation of such mobile infrastructures are the handling of large and variable numbers of wireless connections to the monitored mobile nodes regardless of their current use or locations, and to automatically adapt to variations in the volume of the mobile data streams. This article describes the design, implementation, and evaluation of an autonomic mechanism for load balancing of mobile data streams. The autonomic capability has been incorporated into a scalable middleware system based on a Data Centric Publish Subscribe approach using the OMG Data Distribution Service (DDS) standard and aimed at real-time and adaptive handling of mobile connectivity and data stream processing for great sets of mobile nodes. A significant amount of evaluation experiments of the proposed infrastructure is presented, reinforcing its viability and the benefits arising from the use of an autonomic approach to handle the requirements of high variability and scalability.