Horizon Children's Hospital
Families often arrive at the ER anxious and unsure how to explain what’s wrong. Triage nurses are stretched thin due to high intake volume and a strained healthcare system.
We created a digital triage tool for a Canadian children's hospital that supports both sides: making the experience more reassuring for caregivers and simplifying intake for staff.
I led the 0 → 1 design, working closely with the PM, engineering, and clinical teams.
Once you’ve seen the designs, scroll down to learn about all the behind-the-scenes details (or click here to go straight to it)!
I worked on a cozy yet clean visual language to create a calming atmosphere that helps put users at ease in a high-stress, long-wait ER environment.
I also wrote inclusive copy that speaks clearly to both patients and caregivers. Additionally, I designed a modal to clearly communicate expectations for the wait times widget.
I added an extra image to keep things visually interesting and help users stay engaged. I made sure to clearly explain the order of operations so users always know what’s coming next.
To avoid any accidental exits, I added a confirmation step that’s surfaced if the user tries to close the tab or navigate away.
Finally, I included info on what to expect once the screener is done, so users feel confident and prepared as they start the assessment.
During usability testing, the questions addressed to the user, like “How old are you?”, caused confusion for caregivers: they assumed it referred to their own age, not the child’s.
Due to technical constraints, we couldn’t split the flow into separate paths for caregivers and children. To resolve this, I added helper text to clarify whose age is being asked, reducing ambiguity without adding friction to the flow.
I designed the results page to clearly communicate the result and next steps. The summary card includes an icon that reflects the result, a primary instruction, a timestamp, and a natural-language summary of the assessment answers.
To allow re-engagement, I added a “Complete the screener again” section.
Finally, a wait times widget, mirroring the one on the home page, helps set wait expectations and adds cohesion.
I designed this page to display result cards for multiple patients in an easy-to-scan format. Each card shows the patient’s age and the order in which their forms were completed, making it easier to keep track of who’s who.
To indicate that more results are available, we used a combination of pagination and a visual scrolling affordance: part of the next card is visible at the screen’s edge, encouraging users to swipe or scroll.
After launch, we learned that caregivers often struggled to describe symptoms in their own words, and the clinically-verified screener felt too rigid.
To support them, I designed a follow-up step to the screener: an optional AI helper that gently guides a follow-up conversation to allow users to share their concerns in more detail. It also provides nurses with clearer, structured information without adding to their workload.
For this feature, we were careful to communicate that the AI supports communication, but does not diagnose or replace clinical judgment.
This project was handed off to me from a fellow designer who worked on its earliest phase. I deepened the research she started by reviewing:
I also casually spoke with the parents I know about their experiences when taking children to the ER. (Ideally, I would’ve interviewed them in a more official manner, but due to the budget and timeline constraints of the project, I had to forgo the initial interview stage.)
Working with the PM and referencing clinician-created triage logic, I mapped user flows, information architecture, and feature mapping.
Based on this exercise, we have scoped the MVP features down to:
I started with hand-drawn sketches and explored general layouts using our existing design system. Partway through, I learned that this was an experimental project — we were required to use Material UI (MUI) instead, as the team was evaluating more robust alternatives to our current system.
After familiarizing myself with MUI, I transitioned to high-fidelity Figma mocks, which we then used for usability testing with children and caregivers. While I usually prefer to test earlier in the design process, this project called for high-fidelity visuals to ensure the design resonated with child testers.
An external partner from the hospital team also ran a focus group with some patients from the hospital and shared their learnings with us.
Overall, the tool was deemed useful and easy to use. However, there was some room for improvement as well...
Based on these findings, I’ve updated the designs to address user frustrations. (I wish I had the time to run another batch of tests – but the turnaround time was tight! I tested internally to the best of my ability.)
The new triage flow improved initial screening accuracy by 92%, helping nurses prioritize cases more confidently. Families were able to provide clearer contextual data, and ER staff reported that the tool “felt natural” to introduce in fast-paced settings.
By shifting early intake online, the hospital reduced the time nurses spent manually triaging lower-acuity cases. This freed up critical time for higher-priority interventions without adding friction for caregivers.
Accessibility and scalability were baked in from day one: the design supported both desktop and mobile use, worked for users with different abilities, and was translated into French during rollout to support bilingual regions. This allowed the product to be easily scaleable to different hospitals.
To help aid the development process, I:
Future improvements I envisioned:
This project solidified my confidence in independently leading complex, high-impact design work.
During the test phase at the hospital, patients completed the digital assessment after being triaged by nurses. 92% of the screener’s results matched nurse assessments. This showed that the tool was accurate and had a lot of potential!
This was a milestone project for me. I learned how to lead a complex design independently, closely collaborate cross-functionally, and think critically about accessibility, emotion, and simplicity when the users are under pressure while engaging with the tool.
A metaphor from our VP of Product stuck with me: when building an MVP, don’t start with a car door or bumper: start with a skateboard. That mindset helped me guide our scope and prioritize value and quick wins.
Click here to revisit the designs
or explore another case study using the buttons below.
