CSCW 2011 Talk on "Activity Analysis"

pIT – The Pervasive Interaction Technology Lab

Activity Analysis – Applying Activity Theory to Analyze Complex Work in Hospitals

Jakob E. Bardram & Afsaneh DoryabIT University of Copenhagen, Denmark

CSCW 2011 – Healthcare


• The issue of how to deal with complex, detailed qualitative data from observational studies is still an open challenge for CSCW practitioners, despite much work in this space.

• Approaches• Ethnomethodology• Activity Theory

• However, Activity Theory has been less operational• often on an overall (organization) level• less relevant in interaction design

Motivation


• Applying Activity Theory to analyze detailed workplace studies• Presenting the method of “Activity Analysis”• Different to “Task Analysis”

• ecological valid, re. contextual constraints• focus on collaboration

• Showing how Activity Analysis can be used in analyzing different types of work • Mobility and Distant Collaboration• Co-located Collaboration

Purpose of this paperActivity Analysis


Outline

•Background, Motivation, and Research Questions•Theoretical Background•Activity Analysis•Empirical Cases

• Case A - Mobility and Distant Collaboration• Case B - Co-located Collaboration in ORs

•Design Implications•Conclusion

Outline of Talk


Theory...


• Human cognition and activity are mutually constituted

• (Human) activity is • directed towards an object• ... which is reflected as a motive• mediated by “artifacts” (physical + psychological)• hierarchical and dynamic

• Human activity is (always!) collaborative• by distributing actions amongst people• aligned by the shared motive (reflecting the object)• mediated by communicative artifacts

Activity Theory in one slide...

S

T

O

Activity

Action

Operation


Activity Analysis


1. Detailed workplace studies• traditional qualitative sociological methods of study,

including participant observation, interviews, artifact studies, and video recording.

1. Analysis based on Activity Theory• activities transcribed, and coded using the Activity

Analysis coding schema

Activity AnalysisTwo main parts


Activity – one particular activity, and the actions, people, artifacts, etc, which are involved. For example, the activity of treating a specific patient.

Person – activities and actions of an individual person. For instance a surgeon.

Place – a specific place or location, and the kinds of activities and actions taking place there. For example the operating room.

Artifact – the use of a specific artifact for mediating activities and actions. For example, a medical record in a hospital.

Activity Analysis – Part IThings to study


Activity – The identification and labeling of each unique activity. According to Activity Theory, an activity is identified based on its unique motive and object of work.

Action – The action performed as part of an activity. Operations – The set of manual (or mental) operations that compose an

action. Also enumerated are the instruments (artifacts) mediating the action and its operations.

Context – The context of the action, including time, place, material, and patient involved in all operations that make up this action.

Actors – The human actors involved in an action. Actions are often conducted by one actor alone, but in cases where people work closely together, more people can cooperate on an action.

Activity Analysis – Part IIActivity Analysis Coding Scheme


Coding Schema


Empirical Cases

Case A: Mobility and Distant Collaboration


The activity analysis is based on a person-oriented record by shadowing and video-recording a senior hematologist for a whole day-shift (i.e. from 07:30 to 15:30)

The Activity Analysis helped identify• how many activities he managed during a work shift; • how frequently he shifted between activities; • his patterns of mobility; • the temporal and collaborative patterns of these activities.

Case A: Mobility and Distant CollaborationWork at a hematology department


Activity Analysis – Case A


• 14 unique activities• 11 patient activities, 3 administrative activities

• 86 activity shifts during a 5 hour shift• Fast pace in activity shifting (multi-tasking) during conferences• Activity is not (always) tied to the physical object of work

• e.g. patient-related activity is only done close to the patient 53% of the time

Case A – Results


• In this case, special emphasis was on understanding issues of;• co-located collaboration, • use of instruments, • coordination with people out-side the OR, • the temporal unfolding of actions during surgery.

• The detailed activity analysis is based on video recordings of 5 different laparoscopic operations at a gastric-surgical department of a large teaching hospital.

Case B: Co-located Collaboration in ORsWork inside the Operating Room


Activity Analysis – Case B


• 36 actions divided between 4-5 actors• 137 (human) operations• 33% collaborative actions• 97% involves a physical instrument• 78% involves more than one instrument

• Actions, surgeries, and order• some actions are in all surgeries – some not• some actions can only happen in a specific phase – some not• some actions come in a specific order – some not

• Coordination• using communicative artifacts• through the shared objects of work (the patient)

Case B – Results


Design Implications


• Designing a electronic medical record (EMR)• extremely fast activity switching• interruptions• collaborative

• Context-aware information retrieval• triggered by the proximity to the patient?• no one-to-one relation between context and patient –

more activities may be relevant

Design Implications IDesigning for the Activity


CLINICAL SURFACES

Source: Jakob E. Bardram, Jonathan Bunde-Pedersen, Afsaneh Doryab & Steffen Sørensen. CLINICAL SURFACES - Activity-Based Computing for Distributed Multi-Display Environments in Hospitals. In Human-Computer Interaction - INTERACT 2009, pages 704-717, 2009.

Source: Jakob E. Bardram, Jonathan Bunde-Pedersen, Afsaneh Doryab & Steffen Sørensen. CLINICAL SURFACES - Activity-Based Computing for Distributed Multi-Display Environments in Hospitals. In Human-Computer Interaction - INTERACT 2009, pages 704-717, 2009.


• There was a close correlation between tools and operations• 97% of all actions involved a specific instrument

• The execution of an activity is contextual• Operations are adapted to the specific flow of the surgery

Design Implications IIDesigning for all three levels


Activity Recognition during Surgery

Source: Jakob E. Bardram, Afsaneh Doryab, Rune M. Jensen, Poul M. Lange, Kristian L. G. Nielsen, and Søren T. Petersen. Phase Recognition during Surgical Procedures using Embedded and Body-worn Sensors. In Proceeding of PerCom 2011.

Source: Jakob E. Bardram, Afsaneh Doryab, Rune M. Jensen, Poul M. Lange, Kristian L. G. Nielsen, and Søren T. Petersen. Phase Recognition during Surgical Procedures using Embedded and Body-worn Sensors. In Proceeding of PerCom 2011.


Conclusion


• This paper has presented the “Activity Analysis” method• the method applies ethnographic field studies by analyzing

observations using Activity Theory• provided two cases demonstrating how to apply this method in a

detailed contextual analysis of medical activities inside a hospital. • We have shown that activity analysis can provide detailed insight into the

constituents of human activity.• We showed how Activity Analysis was applied in the design of two context-

aware systems for use in hospitals.• Based on Activity Analysis, we provided a set of design guidelines, which

should be useful in interaction design.

ConclusionActivity Analysis


Please ask, email me, or check my home page [email protected]/~bardram

pIT Labpit.itu.dk

Questions?

Slides available on slideshare


Activity – Focus the design on the (human) activity. • In this case rather than creating a system that reacts on ‘context’ (e.g., location), create a

system that react on human activity.Levels – Design for all three levels of human activity.

• In this case rather than supporting context-aware systems on an action level of e.g. the ward round, incorporate support for the whole patient treatment.

Context – Take into consideration that all human activity is enacted through operations adjusted to the specific conditions of the real world.

• In this case the operations during surgery are adjusted to specific contingencies arising during the procedure.

Collaboration – Take into consideration that all human activity is collaborative involving both concurrent and conflicting actions

• In this case actions and operations may run concurrently involving several persons in the operating room, and context-aware information retrieval needs to accommodate this.

From Activity Analysis to Systems DesignDesign Guidelines

CSCW 2011 Talk on "Activity Analysis"

Technology

Transcript of CSCW 2011 Talk on "Activity Analysis"