3085

An Extensible Measurement Platform on Mobile Devices for Healthcare Monitoring and Personal Wellness Applications


Personal Wellness Application

Objectives

The application aims to provide wellness information of the subscribed user. Together with a powerful mobile measurement platform, the programmable mobile devices can be used as sensor nodes and location-aware data collection instruments. The collected data can then be processed and utilized in various application domains. In order to have a universal access platform, users are allowed to access his/her wellness information via the mobile devices and on the Internet.

Introduction

It is now common for mobile handheld devices like smartphones and PDAs to have a range of sensing (e.g. accelerometers), localization (GPS) and communication (3G, WiFi, Bluetooth) capabilities, in addition to computing. Given the right architecture, these powerful and programmable mobile devices can be used as sensor nodes and location-aware data collection instruments. The collected data can then be processed and utilized in various application domains.

MobiTeC targets to create a measurement platform for different mobile operating systems, on which monitoring tools are developed for different data streams generated from the user's behavior and daily activities (movement data from accelerometers, acoustic data from microphones, location data from GPS, call durations, etc). The data streams will be mined and a reference ¡°physical activity and lifestyle¡± profile of the user can be deduced based on the multiple data streams collected and processed by the proposed system. The developed profile will be used to support both Mid-to-long term personal wellness monitoring, analysis and trend identification/tracking and Real-time abnormal activity detection and alert generation.

A GUI-based application has been developed to provide mid-to-long term personal wellness monitoring and trend analysis. This tool will alert the user of the mid-to-long term changes in potentially health-impacting developments and in the user's physical activity patterns and/or living habits. The application can be accessible via the Internet as web pages. The following information is available at users' interests.

  • Physical Activity Analysis
    • Based on the analysis algorithm of accelerometer, physical activities such as walking or running can be inferred.
  • Geographical Analysis
    • By applying algorithms with 3G and Wi-Fi, geographical positioning of the mobile devices can be located. Besides, location environment (indoor/outdoor) can also be detected with the help of acoustics information.
  • Social Activity Analysis
    • Users may be interested in their social activity analysis, which may be inferred by phone activities and Bluetooth devices. By aggregating the sensor information against time domain, various phone activities can be summarized and presented to users at their interests.

Algorithmic Design

Various classification algorithms are implemented as part of the mobile monitoring tool, as well as the sensor rule engine operation.

  • In- / Out-door classification
    • This classifier returns data to indicate whether the user is indoor or outdoor. The algorithm takes only GPS signals for classification, working relying on the fact that GPS signal reception is bad or even impossible when the device is indoor.
  • Motion classification
    • The online classification of human motion activity on mobile phones requires a lightweight algorithm to process continuous time series data stream from the accelerometer sensor, due to its limited computation resources compared to a desktop computer. However, data stream contains large amount of raw data that may not be easily understand for analysis. By applying dimensionality reduction technique, it is easier to perform mining and analysis on the data.
  • Transport classification
    • The transport classifier adds to the monitoring tool the capability to distinguish whether the phone user is on a transport vehicle or not. The classifier needs accelerometer data (to be specific, the aforementioned motion classification result), cell-id (from radio sensor) and GPS data (for phone localization). There are four states output from the algorithm, still, walk/run, travel slow, travel fast.

Screen Captures

Personal Wellness Application is built on an architecture that can adopt different screen displays on various devices, from mobiles to desktops.

Mobile Login
Web Login
Mobile Login
Web Login

Regarding different information, we provided different views for uses to grab the necessary information at ease. Below show some screen captures of views for different information:

Map View of Physical Activity
Chart View of Location Information
Detail View of Social Information
Map View of Physical Information
Chart View of Location Information
Detail View of Social Information

There are lots more to explore for this application. For details, please go to the Documentations section and download the documents for reference.







.: TOP :.
Last Updated on October 15 2015.
Copyright © 2015. All Rights Reserved. MobiTeC, The Chinese University of Hong Kong.
Disclaimer