Transcript ahm_poster_votes

ual Organisations for Trials and Epidemiological Studies (VOTE
Overview
Architecture
• VOTES is a pioneering project investigating the application of Grid technology to the
field of clinical trials and epidemiological studies.
• It addresses the issues surrounding life-science studies on a macro scale:
• Portal Server – provides an intuitive and easy-to-use point of access for the CVO.
• Grid Server – provides the grid service interfaces that perform the functions of the
infrastructure (authorization, query submission and database allocation)
• Data Server – provides a connecting node to the database pool, through which the
query is passed. Also connects to databases on other nodes of the CVO.
• Driving Database – heads the cluster of auxiliary databases on a node.
• The schematic diagram below shows how a single node on the CVO looks. The box
marked “Oxford” shows how this would connect to a separate node:
• In terms of e-Science, the emphasis of the project is to create a security oriented
data Grid that links disparate data sources from across multiple domains, in a bid to
gain greater scientific and medical insight using the clinical data available.
• Clinical data can be harnessed more effectively using Grid technologies. Three
areas of clinical trials have been identified as particularly pertinent to the goals of the
VOTES project:
• Patient Recruitment – achieving a higher rate of successful recruitment of
eligible subjects, with more efficient targeting over a wider population.
• Data Collection – over the course of a trial, e.g. the drugs/placebos that
patients are taking and measuring their effects
• Study Management – processes involved in recruitment to ensure for example
that the right people see the right data in the right context
• The central point of establishing collaborative federation of trial data is by creating a
Clinical Virtual Organisation (CVO) of participating institutions:
Clinical Virtual Organisation Framework
Clinical
Virtual
Organisations
Framework
Used
to realise
CVO-1
(e.g. for data
collection)
CVO-2
(e.g. for
recruitment)
LeiNott
GLA
Disease
registries
Distributed Data Framework
• Clinical data can be federated across multiple domains using the Grid
infrastructure described above.
• Access to specific parameters is restricted based on allocated roles within
the VO framework
• From this, SQL relating to this particular query selection is constructed and
executed through the data service.
• The screenshots below show the parameter selection screen on the VOTES
portal, with a privileged user on the left and a user with restricted privileges
on the right:
Hospital
databases
Transfer
Grid
GPs
OX
Technologies
IMP
Clinical trial
data sets
• Globus Toolkit v4.0 – developed by the Globus team at ANL, this is used to
construct the grid server functionality.
• GridSphere v2.1 – an open-source and collaborative development (Jason
Novotny, Michael Russell, Oliver Wehrens) is used to implement the Grid
portal.
• OGSA-DAI v2.2 – developed by EPCC at the University of Edinburgh, this
provides the data service for communicating to database pools.
Security
• Resource-level: Individual DBMS administers local accounts for local
resources
• VO-level: Access Control Matrix, administered from VOTES portal, maps
local schema onto VO-wide schema, then applies authorisation controls
on each field parameter.
Data Sources
• SCI Store (Scottish Care Information) – a batch storage system that
allows hospitals to add a variety of information to be shared across the
community.
• GPASS (General Practice Administration System for Scotland) – the core
IT application used by over 85% of clinicians and general practitioners
involved in primary care across Scotland.
• SMR (Scottish Morbidity Records) – a set of data records relating to all
patients discharged from non-psychiatric and non-obstetric wards in
Scottish hospitals.
Summary
• VOTES aims to provide clinical data federation across multiple domains
through the construction of a Grid infrastructure
• Future goals of the project include VO-administration portals, recruitment
portals, visual medical record data sources, integration of records across
national border domains and access through Shibboleth technologies.
Contact / Further Information
• Website and portal: http://www.nesc.ac.uk/hub/projects/votes
• Email: Prof. Richard Sinnott – [email protected]
Anthony Stell – [email protected]
Oluwafemi Ajayi – [email protected]