Towards Understanding Requirements for eScience: the eDiaMoND case study

1
Towards Understanding Requirements for eScience
the eDiaMoND case study

• Marina Jirotka, Rob Procter, Chris Hinds,
  Catelijne Coopmans, James Soutter and Sharon
  LLoyd

2
Requirements and eScience

• eScience and system development
• eScience another domain for technological
  development
• Previous studies of why system development
  projects fail
• Requirements activities
• Eliciting, analysing and specifying functional
  and non-functional requirements
• Development process and managing requirements
  exercises
• Managing conflict and trade offs
• Range of requirements issues
• Security, visualisation of information, data
  storage management and retrieval, data mining..
• Context of work attempting to support - To
  determine what properties a system should have to
  succeed in the environment in which it will be
  used

3
Who needs Requirements?

• eScience is complex challenging domain
• Supporting large scale collaboration
• Understanding interdisciplinary work
• Context of scientific work, how data generated,
  used and shared
• Professional expertise
• Understanding different types of knowledge
• Knowledge in use, not only as classifications
• Range of participants/stakeholders involved
• Biologists, chemists, health clinicians,
  physicists, zoologists, etc
• Academic, industrial, scientific research
  communities

4
Introduction to e-DiaMoND

• 4.1m budget funded through EPSRC/DTI and IBM SUR
  grant
• 2 year project started December 2002
• Academic and commercial collaborators over 12
  sites
• Deliver prototype to support breast screening in
  UK
• Large distributed database of annotated
  mammograms
• Applications will be developed for
• Teaching and education
• Screening/diagnosis
• Epidemiology
• Ambitious, flagship project, with short
  time-scales

5
eDiaMoND Project Team
St. Georges Hospital (GEO)
Guys Hospital (GUY)
Ardmillan
Churchill Hospital (CHU)
Oxford University
6
Context of Requirements Capture

• Challenging complex domain
• Highly volatile - social and organisational
  issues
• Critical - dealing with peoples health care
• Range of participants/stakeholders involved
• Doctors, nurses, admin, researchers, geneticists,
  epidemiologists, radiologists, radiographers.
• Different stakeholders on project, academic,
  industrial, clinical
• Professional medical expertise
• Understanding knowledge
• Tacit understanding and apprenticeship

7
UK Breast Screening Today
Paper
Began in 1988 Women 50-64 Screened Every 3
Years 1 View/Breast 100 Breast Screening Progr
ammes - Scotland - Wales - Northern Ireland -
England
Film
1,300,000 - Screened in 2001-02 65,000 - Recalled
for Assessment 8,545 Cancers detected 300 -
Lives per year Saved
230 - Radiologists (Double Reading)
Statistics from NHS Cancer Screening web site
8
UK Breast Screening Challenges
Digital
Women 50-70 Screened Every 3 Years 2
Views/Breast Demographic Increase 100
Breast Screening Programmes - Scotland - Wales -
N Ireland - England
Digital
2,000,000 - Screened every Year 120,000 -
Recalled for Assessment 10,000 - Cancers 1,250 -
Lives Saved
230 - Radiologists (Double Reading) 50 -
Workload Increase
9
Areas of Technological Interest

• Non digital films and light boxes - transition to
  digital
• Non standard reporting systems - full integration
• Manual movement of data - Grid
• Reporting difficult - Database and Grid
• Training through mentoring - Computer based
  training via Grid
• Localised epidemiological studies - Database and
  Grid
• CADe not used - enabled by Grid and Digital
  Reading

10
Towards Practice-Centred Requirements Analysis
Iterative Requirements Process

• Understanding of work practices
• interviews, fieldwork and video analysis, design
  workshops, prototyping, user acceptance
• Managing conflict between different stakeholders
• design workshops, user participation, trade off
  concerns of technical partners and other
  stakeholders, user acceptance and expectations
• Understanding of transition from user to system
  requirements
• communication and expression of requirements,
  ethnographers and developers, organisational
  constraints, hierarchy of requirements,
  modelling, prioritisation, quasi-naturalistic
  evaluation of prototypes, blueprints, user
  acceptance

elicitation and verification
analysis and verification
specification and verification
? Understanding of local and organisational
concerns
11
Sharing Data within BSU

• Visibility and accountability of work transformed
• Practical ethical action - safety culture
• Flexible role based access structure

12
Sharing Data Across BSUs

• Allows rapid movement of mammograms and patient
  related data between BSUs
• Distributed reading - maximising use of scarce
  skills
• Double reading, professional judgement and trust
• Need to follow case from beginning to end
• Concern over automated allocation of cases to
  radiologists
• Useful in symptomatic clinics

13
Sharing Data Across Disciplines

• Value of database to epidemiologists
• Orient to ethical concerns
• Impact on eDiaMoND project
• Acquiring information to improve healthcare for
  public interest vs protecting citizens from
  unscrupulous use of personal data
• Anonymisation, consent and confidentiality
• Epidemiologists cannot specify complete data
  requirements
• Research work and scientific publications

14
Lessons Learned

• Without understanding the details and context of
  clinicians work we risk building systems that
  are not fit for purpose
• eHealth
• complex collaborative domain
• diverse range of professional expertise
• volatile organisational issues
• Focussing on work practices, organisational
  issues
• Understanding of work flow, collaborative
  practices and everyday work of clinicians
• Techniques to inform design of eScience
  technologies
• Transforming the eScience vision of sharing data
  
• We must be sensitive to the ways in which skills
  such as reading mammograms and researching into
  causes of breast cancer are developed and
  maintained and how these will be transformed by
  eScience technology

15
Lessons Learned 2

• Dont ignore previous research in areas such as
  CSCW regarding global collaboration and virtual
  organisations
• New issues in eScience
• Scale and expertise needed for global
  collaboration - CSCW
• Practical implications from studies of science
  and scientific knowledge - SSK
• Development models and requirements for eScience
  -RE