Transcript CRA-NIH_2006_GreenTeam

Green Group
Jim Foley
John Wooley
Support Previous Discussions; Policy most important
discussions, Recognize Fiscal, Policy Realities
1: Examples of Computer Science Research
results that can help Biomedical Research
• Information guided Collaborative problem solving
(between computer and user) - applied across broad
variety of areas
– Diagnosis, surgery planning / execution
– Smart Decision Support / Computer Aided Human
Insight
• Understanding of underlying computational
complexity of various biomedical problems required
– Appreciate approximations, low resolution theory
to drive experiments: don’t try to precisely solve
intractable problems
– Multi-scale, multi-modal data; integration as life
is, not as academic stovepipes
1: Examples of Computer Science Research
results that can help Biomedical Research continued
• 24x7 biomedical sensors
• Knowledge rep LIMS -> small labs, <$500K
grants. Store, analyze, share data, peer-to-peer
• Quality assurance, error bars, quantity assurance,
“generalizability” commitment from CSE
• Voluntary registration for your data; use of your
data encourages you to put it there. Top down
policy/rules/standards vs. bottom-up www
2: Computer Science Research that would help
Biomedical Research
• Algorithm classification and improvement
• Object recognition, segmentation, registration and tracking
– In vivo microscopy
– Cellular interaction dynamics
– Physiology
• Novel data management systems
– See Jagadish’s article on needs in DB re biomed
– XML Schema for Life Sciences needs (Jim Gray, Microsoft Corp)
• Medical Informatics and current privacy laws – evaluate challenges in
specific context of clinical med; Merging Personal Health record
(digital) to Clinical translational research data. (Complete privacy & security
solutions Take care of HIPPA concerns) EMR problems!
• Ready use” of LIMS by small labs, perhaps unique to NIH Institute
research communities
2: Computer Science Research that would help
Biomedical Research, Continued
• Virtual organizations (VOs) / cyber frontier as the
front-end to integrate teams that relies on
infrastructure, grids, knowledge bases, facilitate
international / large scale studies
• Knowledge representation/ontologies/semantic
web
• Medical informatics for data mining, etc.
• Balance bandwidth and scale
• Visualization of higher dimensional data
#3 Policy Changes to Enable Biomedicine
and Computer Science to Interact
• Biology is complex and quantitative science needs to be engaged
w experiments to ensure impact and acceleration; find cross
cutting fed efforts to encourage collab & impact on education
• Need privacy policy that balances individual policies and public
good – important even if beyond CSE – BioMed discussion
• Interagency cooperation for Cyberinfrastructure
• Data sharing policy, supported by
– Tools & Reward structure that encourages if not requires
– Graded data release, fn of grant size – cf NIH plans
• Specialized centers w more CSE engagement, deal w data deluge
• Reward Structure including provenance, collaboration, sharing,
Federal policy. Journal policy
#3 Policy Changes to Enable Biomedicine
and Computer Science to Interact, continued
• Academic/Government/ Industry/Society Partnerships
– Long Lived Data
– Potential for Commercial / Foundation Contributions
• Encourage / Require computational scientists on study sections and
review panels
• New Pubs or Road show by NIH to CS Depts around country
– Here are our problems, here is how to apply for NIH Funding to work on
them
– Remove misperception “Must have MD as PI to win”
• Distinguish in program announcements and in funding allocations
between:
– Applying known CS research results to biomed research
– Developing new CS research results to support biomed research
– Developing research infrastructure to support biomed research
#3 Policy Changes to Enable Biomedicine and
Computer Science to Interact, continued
– Federal Government recognition, w journals, of importance of
standards ensuring data sharing, maximum impact of
bioscience dollars for research
– Develop capacity of CS trainees into the pipeline using
Training grant mechanisms
• Including summer institutes for Computing professors
– Bring CS and Engineers into the NIH circle. Bring in
Engineers as “partners” in Collaborative research
• Moving further from reductionist science to allow for these other
communities to participate; need balance and sustain road map and
larger collaborations, funding for centers addressing biology challenges
• HCI and Human centered computing community people need to be
brought into the biomedical research community. Change study
sections, CS CO-PIs of PIs, Team requirement.
#3 Policy Changes to Enable Biomedicine and
Computer Science to Interact, continued
• Encourage efforts to bring in quantitative sciences toward
changing nature of biomedical enterprise, from research groups to
NIH. Includes support for NIH discussion of evaluating study sections,
multiple PIs and support for multi-institutional collaborations and
multidisciplinary training; also education challenges, curricula, content – NIHNSF IGERTS on compute – biomed interface
• Cost of infrastructure
– Need to drive Role of infrastructure
• Academic / gov’t / Industry / Society Partnerships
– Community role, federal role
– Identify size, problem
• Feds: data standards and bio; data quality/error bars
– Fund community process to develop needed standards
Policy Changes to Enable Biomedicine and
Computer Science to Interact, continued
• Top down needed: accelerate growth of bio and clinical med
• Require interdisciplinary team of PIs on certain types of proposals
• Another check box, like first time investigator, to give special
allowances to CS and IS investigators special dispensation by study
sections regarding priority score
• Don’t underestimate 2 decades of progress in bringing computing
into biology; don’t be defensive – now speak about success (e.g.,
Eric Jacobson article re top ten contributions, in BCR), but now
accelerate involvement of methods, philosophy and people from
CSE, Math & Stat / physical and computational sciences into
biomedical sciences. Don’t underestimate fact Berg & NIGMS,
Marron & NCRR, plus NIBIB, etc. presence at this meeting.
• Support long term community engagement, not just white paper(s)
in order to encourage cross agency, changes in biomed world and
NIH; partnerships. Professional Society engagement. Can
encourage trans-NIH processes, although centers could be any
Institute.