Development and evaluation of a Behavioural Intervention Grid: LifeGuide

1
Development and evaluation of a Behavioural
Intervention GridLifeGuide

• The LifeGuide team
• Social scientists Lucy Yardley, Susan Michie,
  Judith Joseph, Leanne Morrison
• Computer scientists Dave de Roure, Gary Wills,
  Mark Weal, Jonathon Hare

2
Q What are behavioural interventions?A
Packages of advice and support for behaviour
change

• eat a healthy diet
• cope with illness
• learn parenting skills
• use less energy at home
• work more efficiently

3
Traditional methods of delivering behavioural
interventions

• a) face-to-face, e.g. teacher, therapist, manager
• expert, personalised
• effective but resource intensive
• b) print format, e.g. leaflet
• generic, no support
• cheap but low impact

4
Core components of effectivebehavioural
interventions

• 1. Delivering advice, tailored
• use diagnostic questions to select relevant
  advice from extensive expert resources
• 2. Providing longitudinal support, e.g.
• plans, reminders
• progress monitoring
• progress-relevant feedback
• social support (therapist, peers)

5
(No Transcript)
6
(No Transcript)
7
Advantages of internet-delivered interventions

• For intervention delivery include
• Cost-effective interactivity -- can automate
  delivery of tailored, personalised advice and
  support to huge population
• Access/reach -- available 24/7, accessible from
  home/work/mobile, anonymous
• For intervention assessment include
• Unobtrusive detailed assessment -- can easily
  collect longitudinal data on use/effects of each
  intervention component in large samples

8
The need for a Behavioural Intervention Grid

• Currently each internet-delivered behavioural
  intervention is programmed individually, cannot
  then be changed.
• Developing software to allow researchers to
  flexibly create and modify interventions will
• a) have immediate pragmatic benefits
• b) create the potential for more rapid and
  powerful development of behavioural science

9
Pragmatic benefits of a behavioural intervention
grid

• Improve cost-effectiveness of research
  eliminate duplication of programming
• Improve access to internet-based intervention
  research (e.g. to students, junior researchers)
• Speed up modification/evaluation cycle, rapidly
  optimise intervention

10
The social science potential of a Behavioural
Intervention Grid

• Each modification/evaluation cycle tests theories
  on which intervention components are based
• Networks of researchers can
• a) share intervention components
• b) collaboratively collect much larger datasets
  allowing mediation and moderation analyses of
  effects of intervention components

11
Longer-term scientific potential of a
Behavioural Intervention Grid

• Foundation for a future population laboratory
• semantically enriched, adaptive grid using
  automatic data collection to continuously
  model/refine interventions

12
Planned work of this node

• 2 overlapping, staggered phases
• Phase 1 Develop/test software to provide
  tailored advice
• Phase 2 Develop/test software to support
  sustained behaviour
• Co-design and software development
• Pilot evaluation and software modification
• Substantive test collaborative studies
• Dissemination (throughout)

13
Service Oriented Architecture
Portal interface loosely coupled set of web
services permits easy updating and integration
with other web services software
14
Basis for co-design of software

• experience/prior work of team members and expert
  collaborators
• workshops throughout development expert and
  junior researchers, different disciplines
• internet consultations with networks of
  behavioural researchers
• systematic search of literature and internet to
  identify all techniques used for delivery of
  PC-based interventions

15
System development

• Builds on techniques using Web for e-Science
  established through
• - Semantic Grid e.g. use of Resource Description
  Framework
• - Web 2.0 projects like myExperiment
• Builds on grid experience to support data
  capture and analysis
• Informs emerging practice in e-Science
  for everyday research

16
Pilot evaluations of LifeGuide(plus linked
studentship)

• Observational and think-aloud interview studies
  of researchers and lay users experiences
• explore accessibility, usability, credibility,
  satisfaction, impact on behaviour
• Quantitative data from 100 lay users
• patterns of usage
• responses to website questions and advice
• views of website

17
Substantive test of LifeGuide 1.

• Randomise 2000 users to generic vs. tailored
  advice to self-care for colds/flu
• SEM analysis of cognitive mediators of outcome
• Analyse individual differences moderating outcome
  

18
lifeguideonline.org