Healthcare · Service and behavioural design
A service and behavioural design concept for OLVG-Oost that makes patient-monitor profile status visible at the moment nurses can still prevent unnecessary alarms.
Outcome
A tested behavioural loop started a nurse–doctor conversation and set a clear direction for integration into the existing Philips monitor workflow.
Role
Service design lead
Contribution
Problem framing, mixed-method research synthesis, service blueprint, behavioural strategy, physical prototype, ethical direction and test synthesis
Team
Four multidisciplinary designers
Duration
Nine weeks
Partners
OLVG-Oost, Amsterdam · Philips IntelliVue monitor context
Methods
Field observation, interview, literature review, co-creation, service blueprinting, behavioural modelling, physical prototyping and formative testing
01 · The reframe
Emergency nurses work in an environment where a large share of monitor alarms are false or clinically insignificant. Every alarm still demands attention, so repeated non-actionable signals compete with the alarms that matter.
The obvious design response is to improve what happens after an alarm fires: smarter filtering, routing or a new alert channel.
I challenged that starting point.
I wanted to understand what creates so many alarms in the first place. The research pointed to patient-monitor profiles that remain on generic limits, even when those limits do not fit the patient.
An athletic patient with a low resting heart rate can trigger the same threshold as a patient whose condition is deteriorating. Multiply that mismatch across an emergency department and the system produces noise by design.
The leverage point was not the alarm. It was profile selection at admission.
The brief
Reduce cognitive load when alarms fire
The reframe
Help the care team review patient-specific limits at the right moment, without adding another task or alert
02 · Evidence
Healthcare projects can easily become anecdotal. One interview may reveal a real frustration, but it is not enough to justify a system intervention. I organised the research into four source types and connected every major decision back to more than one.
Field observation — how nurses moved between patients, monitors, the central station and competing alarms
Stakeholder input — senior-nurse interview, co-creation and formative feedback
Contextual mapping — floor-plan movement, workflow moments and mood mapping across a shift
Literature — fourteen peer-reviewed sources on alarm fatigue, behaviour, cognitive load and monitoring
Three patterns kept returning
Profile settings were rarely revisited during transfer
Admission offered more workflow space than the alarm moment
Staff relied on patient context and experience because monitor trust had eroded
Evidence
Interpretation
Design decision
Evidence
Profiles were often left generic
Interpretation
False alarms were being produced upstream
Design decision
Intervene when the profile is first set
Evidence
Nurses were already managing multiple signals
Interpretation
A new alert would add switching cost
Design decision
Add visibility, not another notification
Evidence
Profile ownership was unclear
Interpretation
The problem was also organisational
Design decision
Make status shared between nurse and doctor
Evidence
Staff wanted proof that adjustment helped
Interpretation
A prompt alone would not sustain behaviour
Design decision
Show team-level operational feedback
The research did not tell us what to build. It told us where to intervene.
I connected major design decisions back to more than one source. That stopped the team from treating a strong quote, an attractive concept or a familiar technology as sufficient evidence.
03 · Blueprint
I built a service blueprint across the emergency patient journey, from arrival and triage to monitoring, transfer and team learning.
The most useful finding was not a missing interface.
During the acute phase, no one consistently owned profile selection. Nurses could assume the doctor would decide. Doctors could assume the monitor setup was handled by nursing staff. The behaviour fell between two roles.
That changed the project from a reminder problem into a service-coordination problem. The intervention now had to make profile status visible at the point of care and create shared awareness without turning the metric into staff surveillance.
The blueprint became the project’s decision filter.
04 · Decisions
The question was not which idea looked most advanced. It was whether it reduced alarm burden without asking an overloaded nurse to manage another system.
Refusing an idea was part of the design.
05 · Intervention
Motivation and ability already existed. Nurses wanted fewer non-actionable alarms, and the monitor already allowed limits to be changed. What was missing was a timely prompt, visible confirmation, feedback and shared role awareness.
1 · Bedside cue
A physical reminder beside the monitor shows whether a profile has been reviewed. It sits where the action happens and requires no login, account or new screen.
2 · Team-level operational visual
A glanceable view at the nurse station shows which beds and parameters have been reviewed and how alarm trends compare with a baseline. It supports shared awareness without ranking individual staff.
3 · Briefing cards and stickers
A lightweight social layer introduces the behaviour during onboarding and shift briefings. Paper keeps the data footprint small and avoids creating another tracked task.
Each medium was selected because it could remain small.
06 · Physical iteration
The first wooden prototype used a sliding or flipping panel. It showed red and green, but the gesture was too subtle. Repeating the movement quickly became routine and did not create a clear sense of completion.
We replaced the movement with a spring-loaded click, modelled on the satisfying confirmation of a toaster. The larger colour blocks could be read from a distance. The click confirmed that the profile had been reviewed.
The problem was not visibility. It was feedback.
Click to confirm the profile has been reviewed.
A screen would suggest new input. A click confirms an action that already happened.
The physical object was a behavioural prototype, not the intended final product.
07 · Feedback
The first feedback concept used a face that became happier or sadder depending on alarm volume. It was readable and visually engaging, but the framing was wrong.
A nurse could work through a difficult shift, make sound clinical decisions and still be met by a disappointed face. The system would turn operational conditions into an emotional verdict.
I challenged that direction during critique and co-creation. The team changed the visual from emotional feedback to operational feedback: beds reviewed, parameters reviewed, baseline alarms, alarm trend and previous-shift context.
The visual no longer told staff how to feel about the shift. It gave them information they could act on.
This could feel judgemental during a stressful shift
Emotional evaluation was unsafe and could create blame
Show operational status and trend
Same intention. Different meaning.
The visual direction was developed by the team. My contribution was the behavioural framing, critique and decision to move away from emotional evaluation.
08 · Testing
We tested the concept before it looked finished, using the physical cue, monitor simulation and team visual.
Finding 1
“Visible status started a conversation between a nurse and a doctor about whether the patient's limits should be adjusted.”
The status could support shared ownership, not only remind one nurse.
Product change
Keep the status visible to both roles.
Finding 2
“If I already set it on the monitor, why should I update this again?”
A separate object was acceptable for testing but not for daily adoption.
Product change
Generate the status automatically from monitor settings.
Finding 3
“I do not change every limit at the same time.”
A patient cannot be represented as simply personalised or not personalised.
Product change
Show status per parameter: HR, SpO₂, BP and respiration.
Finding 4
“A total count does not tell me where to go.”
Team totals were informative but not actionable.
Product change
Add bed-level detail so staff can identify where follow-up is needed.
The strongest result was not that participants liked the prototype. It was that the test told us what the product had to become.
Future product direction
The test pointed toward integration with the existing Philips monitor and central-station workflow.
A credible next version would read profile and alarm-limit status automatically, show status beside the bed, show bed- and parameter-level status at the central station, remove manual duplicate updates and keep feedback at team level.
The physical cue remains useful as a pilot tool because it makes the behaviour testable without changing clinical software. The product direction is for the object to disappear while the behaviour remains.
Product hypothesis
Alarmease could add a workflow-level status signal to the existing monitor ecosystem: which beds and parameters still use generic settings, and where review may be needed.
Scope
This is a product hypothesis, not a measured commercial outcome. It requires technical validation, clinical governance and a controlled pilot with OLVG and Philips.
10 · Contribution
I did not lead the final visual design.
My strongest contribution was the reasoning that allowed the team to build the right type of intervention.
I challenged the alarm-centred brief, structured mixed research into one evidence map, built the service blueprint, identified the ownership gap at admission, framed profile setting as a behaviour, co-built and iterated the physical cue, connected ethical risks to concrete design decisions and translated testing into a monitor-integration direction.
My business background helped me listen for the problem behind a stakeholder's first request, structure conflicting information and make the value of a direction clear to different people.
The design skill is not only generating options. It is creating enough clarity that a team can stop pursuing the wrong ones.
11 · Team reflection
The team stayed in exploration too long and ownership became explicit only when we started splitting the prototypes.
I responded by taking a strong role in framing, but framing alone does not create clear collaboration.
Next time I would create decision clarity at kickoff.
Working agreement
Shared goal, quality bar, communication rhythm and definition of done
DACI
One driver and one approver for each major decision
Weekly review
Evidence, interpretation, proposed decision and unresolved risk
Daily prototype check
Dependencies, duplicate work and decisions needing escalation
The lesson is not that I should take more control. It is that I should create the structure in which shared ownership can happen earlier.
12 · Reflection
Before this project, I often started with an artefact, refined it and then gathered evidence to defend it.
Alarmease changed the order.
1. Frame the leverage point
2. Structure evidence from different sources
3. Define what the intervention must not do
4. Build the smallest form that can test the behaviour
5. Test it in context before it looks finished
6. Let the evidence change the medium
GreenPixel showed that I can explore a broad problem, research a market and build a strong concept. Alarmease shows a more disciplined version of that strength: finding the point in a system where a small design decision can change a larger service.
At a glance
Research
Design
Outcome
Scope and attribution
Alarmease is a tested concept, not a deployed medical product. No reduction in alarm volume, patient-safety improvement or commercial result was measured. Prototype data was illustrative and no real patient data was used.
Team: Andyka Jonathan, Barbora Halanová, Negin Bokaei and Duco Boomsma. Duco led framing, research structure, the service blueprint, the physical interaction and ethical direction. The final visual direction was developed by the team.
Next project
In progressCurious about the thinking behind this? Read the bio or get in touch.