First Concertation Meeting BMI Workshop eHealth FP6

blabbingunequaledΤεχνίτη Νοημοσύνη και Ρομποτική

24 Οκτ 2013 (πριν από 3 χρόνια και 9 μήνες)

72 εμφανίσεις

Concertation Meeting, BMI Workshop Brussels 2004
-
03
-
17/18/19

SemanticMining No.507505

First Concertation Meeting

BMI Workshop

eHealth FP6

NoE No. 507505

Semantic Interoperability and Data Mining in
Biomedicine [SemanticMining]


Coordinator


Hans Åhlfeldt
-

Linköping University, Sweden



...
Research

Areas

Health Care

Bioinformatics

Health Statistics


Knowledge engineering


Ontology engineering


Coding, indexing and
information retrieval


Data mining, knowledge
extraction and representation


Natural Language Processing


The Semantic Web


…to support application areas


Information and decision support


Infrastructure for health care
information systems

Application

Areas
...

Concertation Meeting, BMI Workshop Brussels 2004
-
03
-
17/18/19

SemanticMining No.507505

Research focus


Network is about harmonising
knowledge representation
strategies and ontologies


Ontology and knowledge queries
pervade all levels of semantic
web and middleware architectures


Can’t take result of one query,
and use it as argument in next
query to a different service, if KR
strategies and ontologies are
inconsistent, or not compatible


Domain
Ontologies

Information
Models

Decision
Support
Models

NLP
Ontologies

Applications

Concertation Meeting, BMI Workshop Brussels 2004
-
03
-
17/18/19

SemanticMining No.507505

Integration


… to bridge gaps in the European research
infrastructure and to facilitate cross
-
fertilisation
between disciplines …


Computer science (computer linguistics, natural
language processing etc.) [6 partners]


Bio
-

and medical informatics [11 partners]


Health care organisations, standardisation bodies

[6 partners]


SMEs [2 partners]


Concertation Meeting, BMI Workshop Brussels 2004
-
03
-
17/18/19

SemanticMining No.507505

Medium
-
sized NoE


25 partners


11 countries


~ 100 researchers


~ 35 PhD students

Concertation Meeting, BMI Workshop Brussels 2004
-
03
-
17/18/19

SemanticMining No.507505

Partners


N

Short name

Participant

1a

LIU (IMT)

Linköpings universitet, Dept Biomedical Engineering/Medical Informatics,
Sweden


1b

LIU (IDA)

Linköpings universitet, Dept of Computer Science,
Sweden


1c

LIU (C
-
NPU)

Linköpings universitet, C
-
NPU Int.Federation for Clinical Chemistry and Laboratory Medicine

3

KI

Karolinska Institutet, Stockholm,
Sweden


4

SU

Västra Götalands läns landsting, Sahlgrenska University Hospital, Göteborg,
Sweden

5

UGOT

Göteborgs Universitet, Dept of Swedish,
Sweden

6

UKLFR

Universitätsklinikum Freiburg,
Germany

7

UNIFR

Albert
-
Ludwigs
-
Universität, Computer Linguistics, Freiburg,
Germany

8

IFOMIS

Universität Leipzig, Institute of Formal Ontology and Medical Information Science,
Germany

9

CAU

Christian
-
Albrechts
-
Universität zu Kiel, Institut Informatik und Praktische Mathematik,
Germany

10

DIM

Université de Genève, Medical Informatics Division, University Hospital of Geneva,
Switzerland

11

UOM

The Victoria University of Manchester, Dept Computer Science,
UK

12

UCL

University College London, Centre for Health Informatics and Multiprofessional Education,
UK

13

ITRI

University of Brighton, The Information Technology Research Institute,
UK

14

INSERM

Public Health and Medical Informatics Laboratory, UFR Broussais, Paris,
France

15

CNR
-
ISTC

Institute of Cognitive Science, Laboratory for Applied Ontology, Roma,
Italy


16

EMBL
-
EBI

European Molecular Biology Laboratory, European Bioinformatics Institute,
UK

17

MEDINFO

National Institute and Library for Health Information, Budapest,
Hungary

18

NORDCLASS

WHO Collaborating Centre for Classification of Diseases in the
Nordic countries

19

SOS

The National Board of Health and Welfare, Stockholm,
Sweden

20

STAKES

National Research and Development Centre for Welfare and Health,
Finland

21

KITH

KITH AS, Norwegian Centre for Medical Informatics,
Norway

22

NBH

The National Board of Health,
Denmark


24

MRI

Merrall
-
Ross International Ltd, Cheshire,
UK

25

EDSA

European Dynamics,
Greece


Concertation Meeting, BMI Workshop Brussels 2004
-
03
-
17/18/19

SemanticMining No.507505

Budget first year

Concertation Meeting, BMI Workshop Brussels 2004
-
03
-
17/18/19

SemanticMining No.507505

Types of activities

Concertation Meeting, BMI Workshop Brussels 2004
-
03
-
17/18/19

SemanticMining No.507505

Research Areas


Ontology engineering (e.g. anatomy, GO)


SNOMED CT as reference terminology


Concept systems in laboratory medicine


Multi
-
lingual medical dictionaries


Data/text mining


The semantic
-
based electronic health record


Health statistics


Concertation Meeting, BMI Workshop Brussels 2004
-
03
-
17/18/19

SemanticMining No.507505

Long
-
term Commitment


An evolutionary process into a self
-
sustainable network


Key issues for success …


linking to existing networks


creating a momentum


spreading of excellence


fostering partnership, sharing of PhD students, researchers,
infrastructure


industrial and national support


long
-
term co
-
operation among partners, between universities and
health care organisations will follow


successful research will generate new funding


Strategic planning


Network Integration Committee, Scientific Advisory Committee

Concertation Meeting, BMI Workshop Brussels 2004
-
03
-
17/18/19

SemanticMining No.507505

Dissemination of Outcome


Contribution to standards (
CEN TC251, HL7, W3C, ISO)


Educational material


Collaboration with public health care organisations


Disseminate to health care workers & the public


Open workshops and summer schools


Technology transfer


Benchmarks and contests
(
e.g. BioCreative, TREC Genomic Track
)


Industrial liaison


Open source licensing
(e.g. OpenEHR, OpenGALEN, CLEF)


Scientific reporting

Concertation Meeting, BMI Workshop Brussels 2004
-
03
-
17/18/19

SemanticMining No.507505

Contribution to Standards


CEN / ISO


Gunnar Klein (TC 251 Chairman )


Anders Thurin (Vocabulary for Terminological Systems Project leader )


Magnus Fågelberg (European Terminology Group Convenor )


Dipak Kalra (TC 251)


IUPAC


Urban Forsum (C
-
NPU, IFCC
-
IUPAC Chair)


HL7


Dipak Kalra
-

Electronic Health Records


Jeremy Rogers, Alan Rector


vocabulary, terminology


W3C / Semantic Web / OWL


Robert Stevens, Jeremy Rogers


SWISS
-
PROT, Gene Ontology de facto standards


EBI, Midori Harris


OMG Life Sciences Research Domain Task Force


EBI


Concertation Meeting, BMI Workshop Brussels 2004
-
03
-
17/18/19

SemanticMining No.507505

Management


Decision making


The Assembly (all members)


The Board (elected annually by Assembly)


Flexible steering


Committees, Work package groups, Annual budgeting


Communication


WP1
-
9, groupware WP3


Management Office


Co
-
ordinator’s Office
(Hans Åhlfeldt)


International Office Director


Legal Advisor
(Göran Hessling)


Management Advisory Group


Concertation Meeting, BMI Workshop Brussels 2004
-
03
-
17/18/19

SemanticMining No.507505

Concertation Meeting, BMI Workshop Brussels 2004
-
03
-
17/18/19

SemanticMining No.507505

Potential GRID Applications

+

-

N

E

E

D

S


Data Mining



Text Mining and Information Retrieval



Image Processing

Concertation Meeting, BMI Workshop Brussels 2004
-
03
-
17/18/19

SemanticMining No.507505

Data Mining



Machine Learning Approaches



Linear complexity (computer cheap):


naive Bayes, k
-
nearest neighbors



Quadratic complexity (computer expensive):


Support Vector Machine, Principal Component Analysis…



Feature selection algorithms (chi
-
2, mutual
information) might use GRID architecture for
scalability

Concertation Meeting, BMI Workshop Brussels 2004
-
03
-
17/18/19

SemanticMining No.507505

Data Mining and Information Retrieval


Text ~ Data Mining + Data Storage


Inverted Files grows exponentially with the size of the
collection to be indexed


Needed to avoid expensive hardware solutions
(Google)


Example


Indexing 500.000 MEDLINE abstracts = 3 weeks
(using the easyIR tool)


Concertation Meeting, BMI Workshop Brussels 2004
-
03
-
17/18/19

SemanticMining No.507505

Image Processing


Image ~ Data + Text + Storage Explosion


Improvement of peak performance


Performance needs are rare, so harware solutions are overkill



Example:


indexing 5000 radiology images (using GNU
-
GIFT)


5000 images = 40 h on a single machine


Three million pictures are stored in the PACS !



Concertation Meeting, BMI Workshop Brussels 2004
-
03
-
17/18/19

SemanticMining No.507505

Potential GRID Applications


GRID
-
based NLP techniques for information
extraction on free text from clinical reports


high security and authentication requirements due to
sensitive patient data


GRID
-
computing for large
-
scale lexical and
domain knowledge acquisition from Web
-
based
sources


GRID
-
computing for computer expensive
knowledge and language processing applications
(e.g. parsing, terminological reasoning)

Concertation Meeting, BMI Workshop Brussels 2004
-
03
-
17/18/19

SemanticMining No.507505