HEALTHGRID.ORG The European HealthGrid Initiative Grid-based E-health projects in the European Union Tony Solomonides on behalf of The HealthGrid Association
HEALTHGRID.ORG 2 BioGrid 2004, Chicago History DG Information Society Technologies funds projects brings them together in Sept 02 supports conference call HealthGrid 2003 – Lyon, Jan 03 HealthGrid Association incorporated HealthGrid 2004 – Clermont-Ferrand, Jan 04 HealthGrid 2005 – Oxford, Apr 05
HEALTHGRID.ORG 3 BioGrid 2004, Chicago Issues and Projects What is a health grid? combines Grid and e-Science ideas oriented to biomedical advances supports evidence-based practice What projects e-Science led to Grid projects Grid projects included applications e-Health applications took off
HEALTHGRID.ORG 4 BioGrid 2004, Chicago Principal Themes Integration Levels of life & knowledge Correspondence & convergence Issues of ethics and trust
HEALTHGRID.ORG 5 BioGrid 2004, Chicago Principal Themes Integration medical and bioinformatics patient information and medical knowledge ‘gold standard’ evidence and practice- based evidence patient data – images, labs, history trust and ethical issues
HEALTHGRID.ORG 6 BioGrid 2004, Chicago An example Legal Aspects Simulation & Imaging Grid Software Bio-numeric modelling Medical Expertise Secure & lawful Grid provision of medical services Build 6 Grid-enabled medical prototype applications Build suitable middleware on top of common standards Install and evaluate a GEMSS test-bed Anticipate privacy, security & other legal concerns Taken from Gerhard Engelbrecht
HEALTHGRID.ORG 7 BioGrid 2004, Chicago Applications NameDomainClass Maxillofacial surgery simulation Medicine – pre-surgical planning On demand / distributed supercomputing Neurosurgery support Medicine – intra-operative planning On demand Radiotherapy planning Medicine – Monte Carlo treatment simulation On demand / distributed supercomputing Inhaled drug delivery planning Medicine – air flow dynamics On demand / distributed supercomputing Cardio-vascular system simulation Medicine – blood flow dynamics On demand Advanced image reconstruction Medicine – nuclear / in vivo diagnostics On demand
HEALTHGRID.ORG 8 BioGrid 2004, Chicago Applications Compute-intensive numerical methods parallel MPI codes Finite Element Method (FEM) Computational Fluid Dynamics Monte Carlo Simulation ML-EM Iterative Image Reconstruction Method Data Transfers (few MBs to few GBs) Services composed of multiple application components (workflow) Flexibility – User Interactions Near-realtime requirements QoS support required
HEALTHGRID.ORG 9 BioGrid 2004, Chicago MammoGrid EU project to prefigure a pan- European distributed database of mammographic images using Grid Technologies. Aim: To provide a demonstrator for use in epidemiological studies, quality control and validation of computer aided detection algorithms.
HEALTHGRID.ORG 10 BioGrid 2004, Chicago Why a Mammography Database? Improved reliability of screening and early diagnosis requires: better epidemiological understanding improved diagnostic tools enhanced quality control continuous training efficient management of data and records Need to establish research and training repositories that contain sufficiently large statistical samples: MammoGrid-EU NDMA-US eDIAMonD-UK GPCalma-Italy
HEALTHGRID.ORG 11 BioGrid 2004, Chicago Technologies Mammography SMF™ (Mirada) CADe (CALMA) DICOM (Medical Imaging Standard) Distributed computation CRISTAL Database (CERN/UWE) AliEn GRID (CERN)
HEALTHGRID.ORG 12 BioGrid 2004, Chicago Standard Mammogram Form
HEALTHGRID.ORG 13 BioGrid 2004, Chicago The theory of SMF™ Mirada’s Standard Mammogram Form (SMF™) measures the column of non-fatty tissue between the compression plate and the imaging surface. SMF algorithm models the physics of image formation, including extrafocal radiation, scatter, grid effects, film- screen characteristics, etc. The contribution of the imaging system is factored out. The image is decomposed into fatty tissue and non-fatty tissue. The new representation gives a numerical value for the amount of non-fatty tissue at any point on the image.
HEALTHGRID.ORG 14 BioGrid 2004, Chicago Grid Architecture GRID VPN Workstations MammoGrid Data Udine Oxford CERN MammoGrid Data Cambridge GridBox High Security Level MG W/s (‘‘MAS’’) MG W/s (‘‘MAS’’)
HEALTHGRID.ORG 15 BioGrid 2004, Chicago MammoGrid VO
HEALTHGRID.ORG 16 BioGrid 2004, Chicago MammoGrid VO
HEALTHGRID.ORG 17 BioGrid 2004, Chicago Multilevel VO of VOs
HEALTHGRID.ORG 18 BioGrid 2004, Chicago MammoGrid Queries
HEALTHGRID.ORG 19 BioGrid 2004, Chicago Scope of projects MammoGrideDiamond GP CALMA Telediagnosis√√ Quality Control √√ Epidemiology√√ Algorithm Development √(dm)√(dm)√(CADe) Teaching√
HEALTHGRID.ORG 20 BioGrid 2004, Chicago Next steps Ongoing: set-up collaborations address “digital divide” where possible Short term: join other “synergetic” projects to develop Mammography application inside EGEE Long term: new “Image-Grid” project, extending outside mammography domain address distributed database issues
HEALTHGRID.ORG 21 BioGrid 2004, Chicago Principal Themes Levels of life Levels of knowledge Correspondence and convergence medicine and genomics ‘molecular medicine’ ‘individualized healthcare’
HEALTHGRID.ORG 22 BioGrid 2004, Chicago Principal Themes Molecular and Image-based diagnosis Population Disease Patient Tissue, organ Molecular, genetic Genomic Epidemiology Pharmacogenetics Bioinformatics Medical Imaging Medical Informatics Public Health Informatics INBIOMEDINBIOMED PATHOLOGIESPATHOLOGIES Taken from Fernando Martín-Sánchez
HEALTHGRID.ORG 23 BioGrid 2004, Chicago An example Integration of clinical and genetic info from heterogeneous remote databases A vocabulary server that aims to combine existing terminology systems in Medicine and Genetics Novel framework for clinicians to locate, search, access, retrieve and use genomic information in patient care Taken from Fernando Martín-Sánchez
HEALTHGRID.ORG 24 BioGrid 2004, Chicago
HEALTHGRID.ORG 25 BioGrid 2004, Chicago
HEALTHGRID.ORG 26 BioGrid 2004, Chicago
HEALTHGRID.ORG 27 BioGrid 2004, Chicago
HEALTHGRID.ORG 28 BioGrid 2004, Chicago Focus: the individual INDIVIDUALISED HEALTHCARE MOLECULAR MEDICINE Association Modelling Computation Computational recommendation Patient related data PublicHealth Patient Tissue, organ Cell Molecule PublicHealth Patient Tissue, organ Cell Molecule Sofie Nørager Yves Paindaveine DG INFSO Individualized healthcare requires mixing and analysing information at 5 levels: - molecule - cell - tissue - patient - population
HEALTHGRID.ORG 29 BioGrid 2004, Chicago Issues and Projects EUROGRID DATAGRID DAMIEN CROSSGRID GRIDSTART Only health related Grids Multidisciplinary Grids GEMSS HEALTHGRID CLUSTER BIOGRID MAMMOGRID Public Health Patient Tissue, organ Cell Molecule Public health informatics Medical Informatics Medical Imaging Bio- informatics Figure 1 ETC.
HEALTHGRID.ORG 30 BioGrid 2004, Chicago Principal Themes Issues of ethics and trust use of data in care and in research data provenance privacy / confidentiality security national / EU legal framework … extending the concept of a ‘virtual organization’ & building negotiation into the infrastructure
HEALTHGRID.ORG 31 BioGrid 2004, Chicago GRIDs and Privacy Good reasons for talking about privacy here and now! –The HealthGrid promises access to large amounts of heterogeneous distributed data –Health related information is very sensitive and prone to abuse –Privacy impacts society as a whole (e.g. loan applications, insurance, scholarship,...) –Privacy violation is irreversible –Confidential information can never be considered confidential again, once it was out in the open –Grid and Privacy Enhancing Technology exist and are used They could both benefit from early integration Taken from Georges de Moor and Brecht Claerhout
HEALTHGRID.ORG 32 BioGrid 2004, Chicago Available PET Technology Example Technology (PETs for Database Protection): –“Hard” de-identification by the data owner –Anonymisation and Pseudonymisation Techniques –Privacy Risk Assessment –Data Flow Segmentation –Generalisation –Privacy Enhancing Database Agents –Controlled Database Alteration Beyond Technology: –CEN/TC251 standardization effort on AURTAF (Anonymity User Requirements for Trusted Anonymisation Facilities) –CEN/ISSS Focus Group
HEALTHGRID.ORG 33 BioGrid 2004, Chicago Example Application Sharing research data (e.g. disease traceability studies): –Different sites maintain databases: –Nominative records, through direct patient contacts in healthcare delivery –Privacy protected databases (research centers doing data collection) These data could be shared amongst researchers on the Grid, enabling: –Larger geographical coverage (Resource locating) –Standardized data exchange
HEALTHGRID.ORG 34 BioGrid 2004, Chicago Sharing of Healthcare Data Local Databases: –Nominative records (e.g. patient treatment) –Privacy protected DBs Data Access through the Grid Privacy Protecting Interface Privacy Protecting Interface (PETs): –Locally controlled –Pseudonymisation –Content filtering and transformation –Query evaluation (restriction) Local Database
HEALTHGRID.ORG 35 BioGrid 2004, Chicago Privacy Protection Technology Security and Privacy: –PETs and Access Control are complementary –“Security through Privacy” instead of “Privacy through Security” –Within possibilities of today’s PET technology –Knowledge on local data is needed for efficient Privacy Risk Assessment and configuration of PET measures –Data provider is keeping the control: Technically Enforced (= PETs) Dynamic Privacy Policy
HEALTHGRID.ORG 36 BioGrid 2004, Chicago Integrating Grid and PETs Should these privacy protecting services be integrated into the HealthGrid itself? –In the example application, Privacy Protection could easily be separated from the Grid middleware layer (actually such services are already in use) –But there could be considerable advantages if integrated: –Privacy (for databases) policy management and advertising –Synergy of PET and Access Control technology –Could lead to harmonized and standardized PET implementations
HEALTHGRID.ORG 37 BioGrid 2004, Chicago Integrating Grid and PETs Furthermore… –Small “cells” (e.g. geographical area, hospital, …) of anonymous data can lead to increased re- identification risk (i.e. privacy risk) –A “virtual database service” combining several databases through distributed query techniques, solves this –Giving the illusion to the user that a single database is accessed –Provided through Trusted Third Parties (Privacy Policy Enforcing) Virtual databases, policy advertising, … are Grid topics
HEALTHGRID.ORG 38 BioGrid 2004, Chicago Conclusions Summarized: –There is a need for Privacy Protection in a HealthGrid –PET Technology is ready for application and/or integration –The Grid could benefit from integration of a selected number of Privacy Protection Services (in synergy with Security standards?) –Basically, it comes down to selecting an appropriate Grid application, and merging two existing knowledge domains When it comes down to privacy and medical information: Informed consent or other legal measures should not be considered a valid substitute to technical privacy protection techniques!
HEALTHGRID.ORG 39 BioGrid 2004, Chicago More information Web refs HG2004 Proceedings ‘White paper’ to appear June 2004