Transcript The Future of Mobile E-Health Application Development

The Future of Mobile E-Health
Application Development
Exploring HTML5 for Context-aware Diabetes Monitoring
Speaker: Nishant Chettri
Introduction
 Context aware mobile application assisting with self health care
 Digitize manual labor
 Analyze captured data
 Provide personalized assistance
Problem
 Multivariate data input (date, time, glucose level, etc.)
 Lack of good data visualization for decision support
 Cross-platform deployment
 Cost effective
Identified Solution
 Gartner predicts HTML5 web applications as the future
 Cross platform development tools
 Phonegap
 CSS
 Javascript
jQuery UI
Motivation
Context aware cross platform implementation
Mobile
Desktop
Proper data visualization for decision support
Data Visualization
Related Work
 Feasibility exploration on HTML5 for Mobile (Juntunen et al.)
 Performance and Cost/Cross-platform trade off
 Security problems in HTML5 (De Ryck et al.)
 Web Storage privacy, security and performance concerns (West et
al.)
Context Awareness
 Time and Location awareness
 Activity Recognition
 Tri-Axial Accelerometer
 Low pass filter: 5Hz
 Magnitude Filter: Orientation
 Peak Filter: Maxima / Minima
 Step Detector
 High-pass filter and Signal Magnitude Area Filter (Fall Detection)
HTML5 Approach
 Time and Location: W3C Geolocation API Specification
 Accelerometer access requires Hybrid approach [HTML5 + Native
Container]
 Visualization of data for Decision Support
 Parallel Coordinates
 D3.js (Data Driven Documents)
Data Visualization with Parallel Coordinates
Feasibility Analysis
 Devices:
 Mobile: ASUS Android Tablet with Chrome and Firefox web browsers
 Desktop: Chrome and Firefox web browsers
 Offline Use
 Performance Evaluation
 Activity Recognition
 Encryption
 User Experience and Data Visualization
Offline Use
Persistent HTML5 Web storage
Local storage and Session storage
Similar to cookies (Key-Value pair)
Limited to 5 MB
Unencrypted
Lack of storage alternatives
Performance Evaluation
 Activity Recognition
 Negligible for simple context processing
 High CPU load drains battery
Performance Evaluation
 Encryption
 AES and PBKDF2
 JSON data store
 Full pseudo-database encryption
User experience and Data Visualization
 Additional Devices: Samsung Galaxy S3/S4 and Google Nexus 4
 High relative error in sampling frequency
 Loss in data accuracy
 Data visualization powerful, requires larger displays
 Web browsers smoother than Phonegap
Conclusion
 Benefits of using HTML5 and related web technology
 Reduced development across multiple platforms
 Feasible for simpler applications
 Limitations
 Encryption and user experience concerns
 Offline data storage
Future
 HTML5 Specification addresses concerns
 Mitigate performance cost by utilizing the cloud
Questions?