Designed a MyChart extension that helps non-native English speakers understand post-discharge instructions through medically accurate translation, plain-language recovery guidance, and in-portal comprehension support.
Time Span: 8 Weeks
Category: Healthcare
Role: Product Designer
CONTEXT
In the US, more than 25 million people have limited English proficiency. For these patients, post-discharge instructions, such as prescriptions, lab results, follow-up guidance arrive in a language they cannot fully read. Errors in understanding those instructions are a leading cause of avoidable readmissions.
PROBLEM
MyChart centralizes medical records, but it was built for English-speaking users. Patients with limited proficiency can access their records, but they cannot meaningfully understand them. Direct translation fails to convey clinical context: knowing a pill should be taken "twice daily" is not the same as knowing what it is for, what it interacts with, or when a side effect becomes an emergency. This gap isn't simply linguistic — it's a comprehension failure that creates real risk.
"It is annoying having to use Google and Google Translate to know how to use my medications."
Four critical questions went unanswered in initial testing: What is this medication for? What conditions should I follow when taking it? What shouldn't it be mixed with? When does a reaction require emergency care?
RESEARCH INSIGHTS
Four principles guided every decision in the extension, each grounded in a specific finding from research:
Translation without trust. Participants re-translated the same sentence multiple times, not because the translation was wrong, but because they had no way to verify it. A feature that looks trustworthy is as important as one that is accurate.
MyChart as a starting point, not a destination. All participants left the app mid-task to use Google or Google Translate. The app was being used only as a data source, not a comprehension tool.
Task completion doesn't equal comprehension. Participants could recite dosage frequencies but couldn't explain what any medication was for, or what to do if symptoms appeared.
Safety knowledge was the biggest gap. No one could answer what symptoms to watch for, what interactions to avoid, or when to seek emergency care. Participants accepted partial understanding as good enough because they had no alternative.
SOLUTION
The core challenge wasn't translation, it was trust. Patients were already translating. What they couldn't do was verify what they read, understand what it meant for their recovery, or know when something required urgent attention.
The extension addresses this by embedding four layers of comprehension support directly inside MyChart: medically accurate translation, plain-language recovery guidance, a built-in medical dictionary, and contextual educational resources that surface at the right moment in recovery. Together they answer the four questions research left unanswered: what is this for, how do I take it, what should I avoid, and when do I call for help.
ENGLISH VERSION
SPANISH VERSION
OUTCOME
Prototype testing revealed that keeping everything inside MyChart mattered as much as the features themselves. Participants who had previously left the app to use external tools stayed within the portal when translation and guidance were embedded directly. Users responded most strongly to the plain-language medication summaries and the in-context dictionary, both of which addressed the trust gap more directly than translation alone.
"I wouldn't have to leave the app anymore to understand what my doctor is telling me."
"This actually explains what the medication does, not just how to take it."
For patients who had accepted incomplete understanding as normal, having reliable answers inside a trusted environment was enough to change their behavior.
REFLECTION
Over 8 weeks, I learned that designing for non-native English speakers isn't primarily a translation problem; it's a trust problem.
The clearest signal came from a participant who re-translated the same sentence several times during testing, not because the translation was wrong, but because she had no way to know whether it was right. That single behavior reframed the entire project. In most UX work, a confusing interface means a frustrated user. Here, it could mean a missed dose or a preventable readmission.
It also reset our ambitions. We stopped trying to design something impressive and started designing something trustworthy. Those aren't the same target, and chasing the wrong one would have produced a polished product that patients still didn't rely on.
MyChart's extension narrows the gap between access and understanding without fully closing it, and that unfinished distance is exactly what I want to keep working on.
