E-Health:
Chances and Challenges of Distributed
Service oriented Architectures
Prof. Christoph Thuemmler, PhD Centre for Applied E-Health Edinburgh Napier University
© 2011
Dr. Lu Fan, PhD
Centre for Distributed Computing & Security Edinburgh Napier University
Mission Statement
• The Centre for Applied E-Health at Edinburgh Napier
University looks into the interactions and
interdependencies between health in a wider context and information communication technologies and
their direct and indirect effects on individuals, industries and society as a whole. The center is involved in the development, evaluation and validation of novel technologies, strategies and
policies related to the field of E-Health on a national and international level.
Agenda
• Care Models
• The DACAR Project
• E-Health Cloud Platform
• Distributed Cloud Computing Ideas
• Structure Change
• Conclusions & Outlook
Ag
en
d
The when and where
care is delivered is changing
• Average Length of Stay is going down,
e.g. England 2005-2009, from 6.5 to 5.8 days per episode • Overall number of episodes have climbed from 13.7 to
16.2 Million from 2005 to 2009 (England) • Re-admission rate is going up
“The number of patients readmitted as emergencies
within 28 days of being discharged rose from 359,719 in 1998-99 to 546,354 in 2007-08”
• The share of 60+ in the overall number of consultations in hospital rose from 41% to 44% from 2005-2009 although they represent only 21% of the population
• Day cases rose from 3.9 to 5.2 Million over the same period (England)
• Outpatient appointments 2009 compared to 2008 up by 11% (England) C ar e Mode ls Wh en & w h er e car e is d eliv er ed
Drivers for Change
• Demographic development
• Financial Constraints
• Societal Change
• Technological Progress
• Availability of New Products
C ar e Mode ls D riv er s fo r ch an ge
Global Shift in Care Models
Specialist Centered Hospital
based Care, 20th Century
Patient GP Hospital
Distributed Patient Centered
New Strategies Required?
Struc tu re C h an ge N ew st ra tegiesChange is Inevitable
We can’t fight tomorrow’s challenges with
yesterday’s strategies.
C ar e Mode ls Cha n ge is in evit ab leHypothesis
C ar e Mode ls H ypo th esiswww.DACAR.org.uk
D A C AR P roj ec t Ov er viewData Capture & Auto-Identification Reference Project (DACAR)
• Funded by UK’s Technology Strategy Board (TSB) and Engineering & Physical Science Research Council (EPSRC), project no. 400092
• Member of IERC, the “European Internet of Things Cluster”
http://www.internet-of-things-research.eu/ D A C AR P roj ec t Ov er view
DACAR – What are we doing?
• Patient and Object tagging with RFID, Barcode, NFC, all
kind of “License Plate Technology”
• Patient and Object Tracking
• Automated Data Capture (who?, what?, when?, where?) –
• Physiological Parameters, Identity verification in theatre, financial controlling in hospitals
• Data analysis D A C AR P roj ec t Ob je ct iv es
DACAR – E-Health Cloud Platform
• Proof of Concept
• To support fast development, integration, and large-scale delivery of E-Health services
• To facilitate the patient-centric care model by
responding to the challenges of fragmentation and decentralisation
• To integrate applications in primary care, secondary care and assisted living
• To facilitate interoperability
• To facilitate PaaS ? (Vendor lock – in!)
D A C AR P roj ec t E-H ea lt h Clou d p la tf o rm
DACAR – E-Health Cloud Platform
D A C AR P roj ec t E-H ea lt h Clou d p la tf o rmDACAR – E-Health Cloud Platform
• HaaS – two data centres in the UK
Flexiscales, Livingston
Chelsea & Westminster Hospital, London E-Health P la tf or m H ar d w ar e as a Se rvic e
DACAR – E-Health Cloud Platform
• IaaS – open-source hypervisors on Linux
E-Health P la tf or m In fr as tru ct u re as a Se rvic e
DACAR – E-Health Cloud Platform
• PaaS – a stack of software components & services
– APIs for secure web services • Transport-level security • Message-level security
– Single Point of Contact (SPoC) • Authentication • Authorisation – Pillar Services • Data Buckets • Identity Mapping • Audit Trail
• Patient-centric Access Control
E-Health P la tf or m Pla tf o rm as a Se rvic e
DACAR – E-Health Cloud Platform
• Single Point of Contact (SPoC)
– Authentication
• Kerberos
• Federated ID providers: OpenID & U-Prove
– Authorisation
• Policy repository: domain ontology & policy rules • Policy Engine
• Service/Data tickets distribution
– Circle of trust
• Peer-to-peer network of multiple SPoCs
E-Health P la tf or m Sin gle P o in t o f Con tact
DACAR – E-Health Cloud Platform
E-Health P la tf or m In fo rma tion sh arin g p o lic y s yn ta xDACAR – E-Health Cloud Platform
E-Health P la tf or m In fo rma tion sh arin g p o lic yDACAR – E-Health Cloud Platform
• Data Buckets
– Core data
• Atomic medical attribute • Value of a data sample
– Meta data • Unit • Owner • Capturer • Device • Location • Time …… E-Health P la tf or m D at a B u ck et s
DACAR – E-Health Cloud Platform
• Pros
– Contextual privacy – Flexibility – Reusability – Scalability – Auditability – Comprehensiveness• Cons
– Large amount of data – Synchronisation – Network latency E-Health P la tf or m D at a B u ck et s
DACAR – E-Health Cloud Platform
• SaaS – A variety of E-Health applications
– Early Warning Score – Personal Health Record – Object Tagging & Tracking – Evidence-based Medicine – Ambient Assisted Living …… E-Health P la tf or m So ft w ar e as a Se rvic e
EWS Case Study
• The traditional Early Warning Score (EWS)
– A medical algorithm widely used by UK hospitals – Labour intensive & prone to mistakes
App lic ations EWS Cas e St u d y
EWS Case Study
• DACAR’s new EWS
– Fully automated data collection, storage, analysis and risk notification
– Non-stop monitoring in real-time
– Cross-platform: desktop application, web application and mobile application
– Patient-centric governance of data security and confidentiality App lic ations EWS Cas e St u d y
EWS Workflow
• Data Mining, A/D conversion
– Using sensors, RFID transponder tags, smart mobile and handheld devices
– Set up by medical staff, or patients themselves
App lic ations EWS W o rk flow
EWS Workflow
• Doctors/Nurses/Patients subscribe to the EWS
application
– Log on via the SPoC
– Set up data sharing policy for EWS to access vital signs
App lic ations EWS W o rk flow
EWS Workflow
• Medical staff set up the monitor & notification service
– Notifications can be pushed to multiple receivers, e.g. GPs, consultants and carers, via various means
App lic ations EWS W o rk flow
EWS Workflow
• Raw data sent to & stored by Data Buckets in the Cloud
– Data samples are streamed to Data Buckets via secure web service
– Each Data Bucket holds data samples for one attribute – Every data sample is described by a rich set of meta-data
App lic ations EWS W o rk flow
EWS Workflow
• EWS service instance processes data & takes necessary
actions in real-time
– Push notifications, advise on treatments, etc.
App lic ations EWS W o rk flow
EWS – continued…
• Stored medical data can power a wide range of
E-Health applications
– Data access rights controlled by patients themselves – Raw data can be aggregated, sanitised and anonymised – Integrate primary care, secondary care and assisted living
App lic ations D at a p o w er ap p lic at ion s
Chances associated with the E-Health Cloud
• Efficiency savings
• Reduction of the Carbon Footprint • Resources just in time
• Rich third party services
• Hardware & software as a service
ROI in the same financial year!!!!!
• Interoperability
• Enables easy and remote use of smart devices • Cloud Federation to improve security
• Enables the Internet of Things
• Empowers patients to monitor the use of their own data Struc tu re C h an ge Fea tu res o f th e e -h ea lt h Clou d
Challenges
• Amount of data grows much quicker than the available bandwidth
• 4G and 1 GB Internet not wide spread
• Concept of content transfer to a central HUB not sustainable
• Cloud Federation just in it’s early days (authentication
• General public not “informed” about the cloud concept
• Responsibility for information, legal concerns (Google Health, MS Health Vault)
Distributed Service oriented Architectures
• Bandwidth midterm insufficient
• Throughput needs to be maximized
• Architecture of the human nervous systems • Decentralized Processing
Principle of Subsidiarity
• Subsidiarity
• From Wikipedia, the free encyclopedia • Not to be confused with Subsidiary.
• This article is about the general principle of subsidiarity, with particular reference to European Union law. For the social teaching of the Popes, see Subsidiarity (Catholicism).
• Subsidiarity is an organizing principle that matters ought to be handled by the smallest, lowest or least centralized competent authority. The Oxford English Dictionary defines subsidiarity as the idea that a central authority should have a subsidiary function, performing only those tasks which cannot be performed effectively at a more immediate or local level. The concept is applicable in the fields of government, political science, cybernetics, management, military
(Mission Command) and, metaphorically, in the distribution of software module responsibilities in object-oriented programming. Subsidiarity is, ideally or in principle, one of the features of federalism, where it asserts the rights of the parts over the whole.
Distributed Service oriented Architectures
• Hierarchical Structure • Principle of Subsidiarity • Federated approach
• Utilizing different physical networks (Intranet, Internet, NFC, Blue Tooth, 3G, 4G) to reduce bandwidth consumption
• Probably re-packing of data
Distributed Service oriented Architectures
Challenges
• Integration of Smart medical devices on local level • Integration of Smart Phones
• Ad hoc change in configuration and the number VMs (horizontal and vertical) without the need to re-boot (Dynamic adaptation)
• Security issues (Federation) • Pervasive adaptation