NCRI Data Sharing Debate

1
NCRI Data Sharing Debate Peter Dukes Lead MRC
Population Research Data Support Service Medical
Research Council
2
Invitation

• To give a funders perspective on
• benefits / drawbacks of data-sharing and then
  initiate a floor discussion around the issues
  raised.
• on data-sharing, and the obligations that fundees
  are increasingly coming under to share their
  data.

3
Benefits

• Assumption Datasharing is a good thing

4
Assumption Datasharing is a good thing

• OECD
• Publicly-funded research data are a public good
  produced in the public interest
• Europe
• European Bioinformatics Institute
• Community curated databases
• European Strategic Forum on Research
  Infrastructure
• Data Deluge
• Curation and Provenance
• Interoperability
• Multi-disciplinarity of research
• Linking of publications to data

5
Assumption Datasharing is a good thing

• UK Research Councils
• NERC Data Grid
• 9 data stewardship centres
• ESRC Economic Social Data Service
• UK Data Archive
• ESRC requires datasets to be offered to UK Data
  Archive
• MRC
• Policy requires datasharing plans
• Population Research Data Support Service
• e-Science and informatics initiatives
• CancerGrid, NeuroPsyGrid, CLEF, Votes
• NCRI Bioinformatics initiative
• NHS
• Connecting for Health
• Research Capability Programme
• UK Pub Med Central

6
Data sharing

• Science, culture and organisation matter

7
Science, culture organisation

• Databases, archives, biobanks, grids
• Funding model for sharing
• User-led oversight
• Performance service quality science
• Trials, cross-sectional longitudinal
• Investment in the cohort
• Set up
• Nurturing
• Effort to collect, clean analyse
• Performance science outputs

Community data services
PI-led population studies
8
Science, culture organisation

• Not collected for research
• NHS clinical management data
• Social services data
• Education
• Data to support service improvement and policy
• Connecting for Health
• e-Health
• Research capability programme
• Quality service data for patient benefit
• Care
• research

Routine data
Researchable routine data
9
e-health related research landscape
Omics data
Patient groups
Cohorts Trials
Bio Banks
NHS Clinical Data
Demographic data
Educational Environmental Social Data
10
Population sciences perspective

• PI-led research

11
What MRC and PIs say
Hierarchy

• MRC Council says
• Data sharing is the norm in the public interest
• Researchers should submit DSP plans as part of
  grant proposals
• We will fund preservation sharing
• Directors PIs say
• The idea is OK, but for many investigators
  restricted access to their data set is their
  intellectual capital
• Not everyone can be a scavenger feeding off the
  work of others

MRC Council
Funding Boards
Unit Directors
Study PIs
Data scientists Managers
12
What Data Managers say

• Hierarchy

• Benefits of a Sharing Support infrastructure
• Preventing loss of knowledge
• Preserving the value of historic investment
• New opportunities for collaboration
• Discovery
• Open up legacy assets
• Networking spread of good practice
• Technical enhancements
• Automation reduced burden
• Up to date guidance
• No need to roll your own
• More effective induction training
• Greater cost effectiveness across MRC
• Confident partnerships NHS, other RCs
• But we need resources and support from the top

MRC Council
Funding Boards
Unit Directors
Study PIs
Data scientists Managers
13
Drawbacks

• Little evidence of real benefits
• New science improved health
• More efficient informatics
• NCRI Bioinformatics can help here
• MRC Data Support Service - document
• Little understanding of the real costs
• Significant heterogeneity of needs
• We dont know all the technical answers
• Weak public engagement
• Weak academic engagement
• Complexity
• Many initiatives stakeholder

14
End
15
Wellcome Trust
CfH RCP
MRCT
NCRI
BIOBANK
AMS
e-Science
DH NIHR
EBI
CLINICAL
GENOMIC
UKCRC
HEIs
GOVERNANCE
INVOLVE
BBSRC
PIAG NIGB
MRC POPULATION-BASED RESEARCH
FSA
ENVIRONMENT
SUS
NERC
NRES
SOCIO ECONOMIC
MoD
OSI
DEFRA
OECD
DFID
DCC
UKDA
CST
RCUK
UKDF
ONS
EPSRC
ESRC