NST Internship Summary Jan Muzik Presentation Outline About
NST Internship Summary Jan Muzik
Presentation Outline �About the internship and NST �Projects I have been involved in �Results in Call. Me. Smart project �Results in Diabetes project �Practical example – a recent experience �Further outlook �Acknowledgement
Internship at NST � 6 months internship at Norwegian Centre for Telemedicine – NST �Summer 2012, spring 2013 �NST is the world's largest center for research and development in telemedicine and e-health �Located in Tromsø, Norther Norway �Founded by University Hospital of NN �Cooperation with University of Tromso
Participation in NST Projects �Call. Me. Smart Project �The Diabetes ICT Health Motivation Project
Call. Me. Smart Project �Context-sensitive systems for mobile communication in hospitals �Design, development and evaluation of a mobile communication infrastructure that balances communication availability and interruptions for hospital worker �My involvement: �Analysis and redesign of current software solution �Design and development of localization service �Evaluation of new software technologies
The Diabetes ICT Health Motivation Project �Long running project is focused on development and evaluation of ICT tools to increase insight of patients suffering diabetes into their disease �Currently focused on development of and Android application and its evaluation �My involvement � Investigation on possibility of integration of Bloo. Tooth glucometers � Preparing of app evaluation in Czech Rep. � Integration of activity trackers � Preparing of a consortium for a grant application
Call. Me. Smart Project
Current CMS Architecture
Identified targets for improvements �Monolithic application �Lack of app layers �Lack of modules �Lack of interfaces �GUI application �Internal structure is hardcoded �Scalability �Redundancy
Proposed solution �David Wheeler: "All problems in computer science can be solved by another level of indirection. . . Except for the problem of too many layers of indirection” �Modularization �Make the system distributed �Redesign of the database model �Generalization of context sources �Formalization of decision rules �Using a modular framework (MEF)
Things to consider �Essentially a multiplatform solution �Linux, Android, Windows �Extendibility – prepare for context sources and decision rules not know at the time Technologies �Standards �Platform independent �Those we have experience with
Paradigms �First develop in technologies you are the most productive and than watch for the bottlenecks and try to optimize them (perhaps using more efficient technologies) �Donald Knuth: Premature optimization is the root of all evil. �Avoid reinventing the wheel. �Determinism – everything must be stored/logged into local the DB
Architecture Goals – Technical Details �Loosely bound components - SOAP Web. Services �CMS Server �Also the CMS Server has internal modular architecture (Java Beans) �Star architecture �Multiple Context Providers �Centralized database contains all the data necessary for context estimation at any moment
Integration testing - Localization �Integration of Blueth localization service (Blue Tooth based) into CMS �Blueth Server first run at CTU in Prague, later moved to NST �Blueth Agents are running on 4 testing laptops at NST �Blueth API (all web-based) �Web Services – SOAP � Automatically generated proxy object in CMS CAS �REST – OData based – great flexibility – URL specified queries, result in XML or JSON �Currently periodic polling �Web. Sockets callback implementation started
Web. Sockets �Full-duplex HTTP communication �Enables building faster, more efficient application �Extends HTTP protocols, uses port 80 – avoids firewall problems �Implementation problems �New technology – still being standardized (HTML 5) �Supported only on IIS 8 – Windows 8, Windows Server 2012 �Support on older Windows platforms would require change in OS core – http. sys (IE 10? )
Context Provider Architecture
Blueth Localization Service �Resolves original problem of high energy consumption for phone invoked Bluetooth localization �Uses either : �Bluetooth device discovery �Station Connect � Method for detecting BT devices even when not discoverable – modern phones have only limited discoverable time � Periodically tries to open BT data channel �Client-Server architecture �Uses newest Microsoft technologies: NET 4. 5, IIS 8, SQL Server 2012, EF 5 �In proposed architecture BT localization can easily be superseded by more precise localization technology (ultrasound)
Blueth Architecture
Blueth Server �Blueth. Hub – IIS hosted, WCF SOAP services �Web. Monitor - ASP. NET web app. �Dynamic. Data – Scaffolding web administration interface (web app) �Data. Services – OData, universal REST web API, URL queries �Microsoft SQL Server 2012 Database
Blueth. Hub API (SOAP)
Blueth Data Services
Data. Services OData �http: //ds. blueth/Data. Services. svc/Are as? User. Id=1&$orderby=Display. Name
Windows Phone 7. 5 �Monitoring of localization process �Prototype application �Based on Silverlight WP 7 app. �Connected via WCF (SOAP) �Web. Sockets not supported on WP 7. 5 - Periodic polling �Built on MVVM architecture pattern �Extensive data binding, WPF �Declarative approach – almost no code
Diabetes Project
Blue. Tooth Glucometer Integration �Only true BT glucometers are made by Fora �Fora G 31 �Fora Diamond �Same communication protocol �Just after the BG measurement device tries to connect to a paired mobile phone �If the connection is successful, measured values can be obtained
Prototype Application �Sugar Monitor �Receives BG data via Blue. Tooth �Sends measured values to a server �Communication module is running as a service in background �Received BG measurements are send to the server via OData service �OData library: k. SOAP 2 �Data can be reviewed only via web portal
Physical Activity Trackers �Energy expenditure is an important value that influences actual BG �Two approaches (or combination): �Continuous long term activity monitoring – pedometers � Fitbit, Nike Fuelband, Jawbone � Must be worn permanently �Only peak activity tracking – mobile applications � Sports. Tracker, Endomondo, Runkeeper � Relies on GPS – works outdoor � Requires manual specification of activity (walking, running, cycling)
Activity Tracker Integration �Most suitable tracker is Fitbit Ultra �Requires specialized app to upload data from the tracker �Supports both Android and i. OS �Data can be obtained from server via web API – Oauth secured �Provides number of steps and calories burned per time period - up to 1 minute granularity � 3 month successful test of 4 users
Why to travel? A practical example �The Time-zone Problem �Telemedicine (IT, distributed systems) related problem �In Europe, not that actual problem �People (telemed. patients) not often cross the time zone boundaries �Daylight saving (letní čas) �Discontinuity in time �Example issue: �Mobile emergency system �Many dropout alarms generated
Holidays in the USA
Time zones in the USA �Mostly 4 time zones… �People are crossing TZ borders frequently
Time zones in the USA �… but Arizona does not observer DTS
Time zones in the USA �…but Arizona’s Navajo Nation DOES observer DTS
Time zones in the USA �…except of Hopi Nation within Navajo reservation!
World Time Zones �Offset time zones: �India: UTC +5 ½ �Australian Central Time: UTC +9 ½ �Nepal: UTC +5 ¾! �Nepal time used to be UTC +5: 40 20 years ago �Etc. �Whole China uses one time zone and no DTS
Solution: Coordinated Universal Time �UTC -> compromise between French Temps Universel Coordonné TUC and English Coordinated Universal Time CUT �Derived from International Atomic Time �Most days have 84600 s, but adds leap seconds (25 in total since 1972) �Best practice: �Storing all times (in database) in UTC �Converting times to local time only when being presented to user �Even MS Windows works this way
Benefits for Students �New Erasmus agreement (2013 -2014) �Possibility to study 1 -2 semesters at Ui. T �EU co-financed stipendium � 30 student-months per academic year (Medical Technology) �All study levels: �Bachelor �Master �Postgradual �Life changing experience! �International Master’s Program in Telemedicine and e. Health at Ui. T
Further Perspective �Formation of a research consortium: �FBME CTU �NST � 1 st Faculty of Medicine CU �Motol University Hospital �Several SMEs �? �Preparing a grant application for Norway funds �Waiting for the call since spring 2012 �Grant proposal focused on Evaluation of telemonitoring motivation of diabetic patients
Acknowledgement �Terje Solvoll �Lorenzo Gironi �Alain Giordanengo �Bernd Talsma �Eirik Arsand �Gunnar Hartvigsen �Rune Pedersen �Tonny Iversen �Yngve Bergheim �Radim Kliment �David Gillar �Miroslav Mužný �Jan Brož �Martina Oulická �Karel Hána �Jan Kašpar �Vojtěch Malina �Eva Caithamlová
Thank you for your attention! Questions?
- Slides: 47