Anaesthesia nurses play a key role in the operating theatre. The anaesthesia machine, however, has grown so complex that many do not use it to its full capability. In this high-pressure environment, nurses have no room to fail so are risk-adverse.

This project explores how nurses could experiment and deepen their knowledge of how the anaesthesia machine affects the patient.


With Sandbox, nurses can see a prediction of the patient’s response to drug parameter adjustments before administering any changes. Through this system, the nurse can explore, fail and learn without risk of harming the patient.

Group project (2017, completed over 10 weeks) at the Umeå Institute of Design in collaboration with Getinge.

Team: Selvi Olgac, Minh-Huy Dang

With Sandbox, nurses can simulate drug adjustments and see an individual prediction of the patient’s response

How might we create a learning environment for exploration during the autopilot phase, without losing focus on the patient?

The perfect team

Sandbox works in collaboration with the nurse, listening to feedback and learning from the patient’s reaction.

Ready to learn

Continually evolving, Sandbox adapts to future changes; no software updates required.

Flexible futures

Once the operation is underway, supervisors can leave trainee nurses to act independently, freeing expert time.

Local control

Hospitals store anonymised patient data locally, removing reliance on third parties and enabling national and international cooperation.

Sandbox: Patient Prediction in the OR


Early Research

Diving in the Deep End


Hospital visits


Surgery observations


Nurse engagements
Understanding anaesthesiologists

Immersing ourselves into the lives of the anaesthesia nurse, we set out to understand how and why they work. Surgery observations allowed for us to see their workflow first-hand, in their context. We used ethnographic methods to explore the complex thoughts and feelings of daily life.

The role of the machine

Establishing the current role of the anaesthesia machine early was crucial to define how it could evolve in the future. As well as nurses walking us through several models, we observed them during operations to learn the real-world patterns.

Creating thick stories 

Mapping insights

Obligatory postit photo


What we learnt from our research

20/80 knowledge

Due to the complexity of the anaesthesia machines and lack of resources for continual training, nurses have limited knowledge of their functional potential.

Double checking

Nurses must continually monitor the patient through data displayed on a screen and visual checks. The OR layout means they cannot monitor both the screen and the patient at the same time, risking information being missed.

Boring autopilot

Whilst the start and end of the operation are high-pressure, the lengthy middle section (dubbed the ‘autopilot phase’) is usually uneventful. The nurse may monitor the patient for over 6 hours, needing to hold focus.


design sprints and concept development

1. Second Skin

How might we get feedback directly from the patient during autopilot?

2. Machine Dependence

How might we maintain the human touch between the nurse and the patient during autopilot?

3. Interactive Sandbox

How might we help nurses to better understand the machine-patient relationship through exploration and failure?

Early concept ideation

Bringing lofi prototypes to users


User Testing

Testing visual design

Defining information hierarchy

Scenario role-playing

What I learnt

I have learnt a huge amount over the 10 weeks of this project, entering an area new to me that is both hugely complex and well established.

The importance of carrying key findings through from research to execution was abundantly clear in our process, as well as how difficult this can be. I enjoyed our thorough approach in which every design decision could be traced back to an insight – and learned how valuable this is for the final outcome.

I was an active team member throughout the project, pushing for a model of ‘togetherness’ where the design process is kept wholly collaborative, with work only being divided in the final period of execution.