User Requirements and Engagement in Health Informatics Alistair Sutcliffe Sarah Thew, Oscar De Bruijn, Manchester Business School, Jock McNaught National.

User Requirements and Engagement in Health Informatics

Alistair Sutcliffe Sarah Thew, Oscar De Bruijn, Manchester Business School,

Jock McNaughtNational Centre for Text Mining

Iain Buchan, Paul JarvisNIBHI, School of Medicine

Rob ProctorE- Science Centre

University of Manchester

ADVISES project Adaptive Visualisation Tools for e-Science Collaboration

EPSRC E-Science Usability  program

The Vision

• Maps and visualisation to support Epidemiology research and PCT analyst- local health policy

• Easy to ask questions (Not SQL) as the user interface

• Interactive Visualisation allows users to see the answers to question

AND explore data in rapid ‘what if’ mode

• Two projects:

Obesity Atlas- mapping tools for epidemiology

ADVISES- more generic mapping and visualisations tools

The Domain- Epidemiology

UnderstandingChildhoodobesity

Causal analysisfrom complexmultivariatespatio- temporalevidence

Multi-variate statistical analyses- differences between cohortsover time, between areas

Interactivevisualisation

See the effects of differentAnalyses- in context (space, time.distribution in population, etc)

Researchquestions

Requirements Analysis- Approach

• Ethnographic studies- observing research practices

• Interviews for background domain knowledge

• Language analysis- analysing published papers and recorded conversations (Research Questions)

• Scenarios and Storyboards- early designs for- Primary Care Trusts- visualisation of epidemiology of childhood obesity - Epidemiology researchers- visualisations of complexdatasets

• Requirements workshops and demonstrations

User Engagement- policies

• User consultation- presentations at PCT working groups- Manchester Obesity task force

• User Centred design- iterative prototyping- evaluation approach

• Participatory design, users involved in taking design decisions via storyboarding sessions

• Long term partnership NIBHI research centre plus outreach mission for local Primary Care Trusts

ADVISES PCT prototype

Analysiscontrols

InteractiveMap display

Multiple representations

Quick win prototype- more complex controls and functions added later

Obesity Atlas prototype

Evaluation

• Regular formative evaluations with users

• Qualitative task based analysis - ‘thinking-aloud’

• 3 rounds of evaluation, 21 participants– 8 male, 13 female– 8 academic epidemiologists, 13 NHS Primary Care Trust

analysts

Results- effective operation over a range of tasks- 92% completed- users positive ratings – mean 82% satisfied over range of Qs

Lessons learned

• Maximise access to user/domain experts- - diversify user base - engage users with storyboards and prototypes early- go with the flow- follow your users’ enthusiasm

• Understanding the domain– background reading– appropriate expertise on the team– talk the users’ language

• Prioritising Requirements- cost/benefit analysis for trade offs- look for quick wins for user engagement

Achievements

• Prototypes of Obesity Atlas and ADVISES complete

• Several iterations of user centred design (Obesity Atlas) and participatory design (ADVISES) improved prototypes and incorporated users design ideas

• Evaluation studies also explored how prototypes could be used in practice – workflow plans

• Roll out of combined prototypes under way- funded by PCT users

• Prototypes adopted and being refined by researcher users