Case study 04 / Enterprise training platform

Scalable training for ICU nurses: no-code scenario builder & game

A mobile training game and a no-code content management system that let non-technical instructors build high-acuity patient scenarios for junior nurses to practice on.

Role
UI/UX designer
Client
A hospital in Singapore
Duration
October '23 – January '24
Focus
User interviews · Sketch prototyping · UI & interaction design
Split view of the no-code scenario builder CMS on a laptop, showing a branching decision flow, next to the mobile training game on a phone, showing a patient's torso with numbered assessment points.
The two halves of the platform: a no-code scenario builder for instructors, and a mobile game for junior nurses.

Challenge

Junior healthcare workers face a steep learning curve in high-acuity care units — resource constraints, theoretical training gaps, and limited clinical practice opportunities. Traditional training is time-intensive and doesn't scale with rising demand, and ineffective learning has a direct, serious impact on patient care.

How might we train junior healthcare workers more efficiently and interactively, so they can identify and manage patients in acute medical conditions?

Results

A training platform comprising a mobile game and a no-code CMS, letting instructors build and update high-acuity scenarios themselves, without engineering support.

200 Average game sessions
73.5% Gameplay improvement
25+ Pilot players

Hiring nurses was never going to be enough on its own.

The attrition rate among nurses has been rising for five years, pushing the nurse-to-patient ratio from 1:3.75 to 1:10. That pushed the government to hire 4,000 nurses in 2023 — but hiring is only part of the solution. Once hired, they still need to be trained, to keep patient care quality intact.

All health professionals, including nurses, don't hit the ground running, and need an appropriate period of supervision and mentoring, the specifics of which would depend on training and prior experience.

Dr Jeremy Lim, public health specialist

This project marked my first foray into game development, and the challenge that came with it: since the end users would be instructors, not developers, the solution had to require zero coding skill to operate.

Getting first-hand experience of the induction program.

To truly understand the end-user's perspective, we immersed ourselves in the ICU department to experience its atmosphere firsthand — starting at the learning centre, then observing live ICU rooms.

🧠

Cognitive overload, both when undergoing training and when handling a live patient case. Beyond operating machinery, nurses had to monitor conditions and make critical decisions at the same time — they had to learn to prioritise and act decisively.

⛔

Lack of practice, caused by limited access and availability. Permissions are required to use essential machines, so junior nurses' opportunities to learn are often opportunistic, not guaranteed.

😖

Disorganised information, taught informally and on a feel basis. Critical information — treatment protocols, expected physiological responses — is scattered across preceptors, with little standardisation.

Together, these insights pointed to a solution that had to be both engaging and accessible — but without a clear, well-organised structure, information would just become another barrier instead of removing one.

Bite-sized, scalable, and visually honest.

Further interviews with nurses surfaced three requirements: bite-sized information that doesn't overload; scalability, so new scenarios can be added and disseminated as training needs evolve; and visual reinforcement, so nurses can see the real impact of their decisions on a patient.

The direction we set

An accessible, bite-sized format that lets nurses practice decisions and see their impact.

The solution also had to let administrators seamlessly add new scenarios themselves, to keep pace with evolving training needs — a requirement that would end up shaping almost every decision from here on.

From a linear template to one that branches and loops.

Once the format was set, my team went deeper into the content itself. There was no existing resource or structure for it, so we gave nurses a template and a sample scenario to work from.

We quickly realised the scenarios were far more complex than we'd anticipated. Clarifying them took extensive back-and-forth — every detail had to be understood as accurately as possible, and every feature we proposed had to stay scalable.

A sample scenario template, 'Content Sample — Chest Pain', broken into Context, Trigger, Decision Making Point, Options, Impact, and Result columns, each populated with in-game screenshots.
The CTDOIR template we handed nurses to structure a scenario — before we knew how complex their answers would get.
Internal team's challenge

Coverage

How could we be sure we had every necessary medication? Could the client even provide a comprehensive list?

Nursing team's challenge

Time

Gathering every piece of medication information was time-consuming — and it was never obvious why something had been missed.

To keep clarifying scenarios like this from turning into a bottleneck, we translated each one into a consistent visual language of dialogue, vital-sign monitors, animations, and pop-ups — something both nurses and developers could read the same way.

Annotated diagram labelling recurring interface patterns in a scenario flow: Dialogue, Vital Sign Monitor, Animation, and Pop-up.
Naming the recurring patterns first made every scenario after this one faster to translate into the system.

An exhaustive list of medications was never going to be possible.

Medication is one of the most common treatments in any scenario, but it carries its own complexity — when and where to administer, dosage, delivery method, limits, increments. As we gathered detail, it became clear that no list of medications we compiled up front would ever be complete.

So instead of trying to enumerate every medication in advance, we built the Button Manager: a panel that lets senior nurses add new medication options themselves, set the correct dosage and status, and attach a reaction — all without waiting on a developer.

The Button Manager panel: fields for Interaction Type, Major Button (Oral), Sub Button (Lisinopril), Correct/Incorrect status, and dosage amount, with a 'Substep added successfully' confirmation toast.
Adding a new medication option — no code, no developer, no wait.
A box of Lisinopril 20mg tablets, the medication used as the example in the Button Manager screenshot.
Lisinopril — the example medication nurses used while testing the Button Manager.

As new scenarios arrive, the medications they need are just as likely to be missing from the system as present. The Button Manager turned that gap from a recurring development request into something senior nurses could close themselves, in minutes.

Observation fosters empathy, and this was a mental gymnastics of its own.

Working on this project gave me a newfound appreciation for nurses. Most people recognise that nursing takes passion and emotional resilience — after our first ICU visit, I realised how much more is involved. Beyond emotional strength, nurses need deep medical knowledge too, something usually associated more with doctors.

This project also pushed my engineering mindset harder than most: extracting patterns from complex information, understanding intricate linkages, and building logical flows that stayed both realistic and scalable. Requirements kept emerging as we tested with the client, and gathering everything upfront was never realistic — so we designed and built in parallel with that uncertainty, rather than waiting for it to resolve.

Overall, this was a genuinely rewarding experience — one that strengthened my ability to empathise with users while sharpening how I reason through complex, ambiguous systems.

End note

Need instructors to build their own content, without a developer on call?