Obesity eLab 20090129 Shoaib Sufi

1
Obesity e-Lab2009-01-29Shoaib Sufi
2
The Obesity Problem

• Obesity is associated with many illnesses and is
  directly related to increased mortality and lower
  life expectancy
• It is commonly defined as a body mass index (BMI)
  of 30 kg/m2 or higher
• In the UK tackling Obesity is a government wide
  priority
• 'The anticipated growth in obesity in almost all
  segments of the population is quite alarming, and
  the analyses of the Health Survey for England
  give a strong impression of continued inexorable
  growth'
• Foresight, Tackling Obesities Future Choices -
  Modelling Future Trends in Obesity Their
  Impacct on Health (2nd edition) Government Office
  for Science
• Not just about Growing rates Epidemiologists,
  Social researchers and Public health officials
  want to understand (and try to control) the
  reasons beyond eating too much and not
  excercising or distal factors which cause an
  increase in Obesity rates. A diagram created from
  a review of outcomes of Foresight Tackling
  Obesities project shows the identified 108
  factors which drive obesity.

3
(No Transcript)
4
How Researchers work
Results
Data
Analysis Models
Publications
Data is Key this is where it starts
5
Digital Dust (data deposit gt use)
Surveys
Research
Clinical
Public Health
Deposit
Use
NHS Partners Data Tombs
6
It starts with Data Health Surveys

• Surveys under used and difficult to manage
• Health Survey for England HSE
• British Household Panel Survey - BHPS
• Within the same survey
• Hard to Navigate
• Cryptic label
• Metadata separate from data (cross-referencing)
• Differences
• Labeling of the same concept year to year
• Differing in measurement technique of the same
  concept from year to year
• value domains of same concept different year to
  year

7
Data Issues continued

• Across Surveys many issues
• Matching concepts across surveys
• Matching value domains
• Combinatorial effect of mapping the whole space
• conceptssurvey yearsnumber of surveys - not
  feasible for just one researcher or group!
• Once portion of interest done not easy to reuse
  by the person themselves (e.g. 6 months later) or
  to share.
• What's needed Just Enough (mapping), Just In Time
  (when you need it) and Just Shareable (to
  co-workers or the community) i.e. Data
  Methodologies

8
Understanding Surveys
9
Data Methodologies
10
Benefits

• Allows sets of mappings to be shared and added to
  allowing collaborative building of Data
  methodologies
• Does not get rid of the problem in surveys but
  allows a mechanism for capturing the
  re-conciliation process in a structured manner so
  it is not ad-hoc or bespoke
• Allow re-discovery by the original author
• Saves time time better spent doing research and
  not data manipulation/harmonisation.

11
After Data ...

• Once you have access to data in a sensible form
  then
• You want to analyse data through various common
  categories and patterns of analysis Analysis
  Methodologies
• You want to visualise the data or the analysis
  common charts spatial-temporal mapping
  Visualisation Methodologies
• Necessary pre-conditions making things explicit
  that are normally implicit. Frames of reference
  or earlier research Reference Methodology.

12
Obesity e-Lab Architecture
User
Research Questions Answers
Obesity e-Lab Workbench
Value
Composing, Sharing, Collaborating
Data Method
Analysis Method
Visualisation Method
Reference Method
Survey xWalks
Analysis Catalogue
Charts, Maps, Output areas ...
PubMed, ...
Data, Surveys, Analysis, GIS, Publications
Survey Providers

13
How e-Labs will help future Obesity analysis

• Sharing types of analysis
• Chaining common analysis
• Shared understanding in research group
• Better publication
• Built to be repeatable
• Types of Analysis (Epidemiology nomenclature)
• Descriptions
• Distributions
• Trends
• Associations
• etc
• Aim is to cover the 20 of the space of
  Epidemiologics/Social Researchers/Public Health
  people that gives 80 of the value/usage.

14
Obesity e-Lab workbench (WiP)
Concepts
Analysis Chaining
Concept Mapping
Sharing
Visualised Results
Building the RO's
15
Clear Public Good
Unclear Public Good
Research Objects
Research
Depersonalise
e-Lab Population
Health Records
Health Records
Local Ownership
Asset Enrichment
16
Obesity e-Lab going forward

• Standing on the shoulders of many Giants
  simultaneously !
• Obesity e-Lab is an instance of an e-Lab
• Built upon same principles
• Notion of RO as common currency
• Linked with like minded project to allow creation
  of inter-operable pieces
• Users and Social researchers are excited about
  the prospect of such a tool
• Easy re-use and re-purposing of work
• Possible use by data providers to create
  definitive concept maps
• Save time get on with research

17
Obesity e-Lab Overview

• Obesity is associated with many illnesses and is
  directly related to increased mortality and lower
  life expectancy.
• Due to the nature of health related survey data
  there is an underuse of large health survey data
  (e.g. The Health Survey for England).
• If access and understanding of information inside
  surveys were made easier the use of such survey
  data could dramatically increaed.
• Obesity e-Lab aims to lower the barrier of entry
  to data, analysis and visualisation around survey
  data, allowing deeper analysis and thus better
  policy decisions with a focus on the Obesity
  problem.
• Obesity e-Lab is an instance of an e-Lab built
  using a common format (Research Objects) and
  service notions.
• Research Objects allow a replayable, shareable,
  descriptive and methodological understanding of
  research outcomes.

18
Questions

• ?

19
Index Slide

• The Obesity Problem
• Current Analysis Problems
• Proposed Solutions
• Why e-Labs are the right approach
• Research Object for Sharing
• Common technology landscape
• How e-Lab will help the future Analysis
• Current Models of Surveys and Methodologies
• Software RoadMap
• Current Screens