Research
Why do people abandon the systems built to support them?
Two years of ethnographic accompaniment across medical, administrative, and care settings. One deployed platform. One PhD question still open.
What looked like non-compliance with medical advice almost always had complex underlying causes. You only saw why if you stayed close enough, long enough, to watch.
Where this started
Two patients, one caregiver, and the gap that built Laguna
I put my screenwriting career in Europe on hold and came home to care full-time for my parents, who were both being treated for cancer at the same time. What I encountered between hospitals was not a technology problem. It was a system that had never given my parents a way to say what they actually needed.
One doctor prescribed seven or eight tablets. Another prescribed a similar amount. The drug interactions left them feeling worse, so they simply stopped taking everything. They refused to exercise, not from stubbornness, but because their hips and knees were painful in ways they had no language to describe to a doctor in a five-minute appointment. The system recorded this as non-compliance. It was not.
The systemic frame
Taiwan's National Health Insurance system is celebrated for coverage. A typical appointment runs five to ten minutes. Electronic health records exist, but data across hospitals often arrives as noise rather than signal.
The deeper problem: the system had never built a mechanism for patients to organise and carry their own experience into the room. What was missing was not information technology. It was a way for a person to be legible to the system without losing their own account of themselves.
This is the question that reshaped what "accessible" means to me: not a standard a space meets, but a condition a person can live inside.
Deployed work
Laguna Companion Service
A patient and caregiver platform for consultation preparation, focus support, and post-visit processing. Built from scratch. Live in pilot since July 2026.
What it does
Laguna lets patients build a running record of how they are doing between visits and uses AI to help them organise what to bring to the doctor. After the consultation, the record and follow-up context can be transferred securely to whoever accompanies the patient next.
laguna.ailunhuang.comBuilt with
Design principles
Mandarin · English · Indonesian · Filipino · Vietnamese
The next question
How does an adaptive, human-centred AI hold onto a person whose motivation, capability, and life circumstances change across months and years?
Laguna is my first real attempt to turn these observations into a working intervention. The live deployment will itself become a site of observation. The PhD question is not how to build this kind of tool, but how it sustains engagement without adding burden or eroding the dignity of self-management.
Methodology
Over two years, across medical, administrative, and transport settings. Not interviewing, but accompanying long enough that people tell you in the third hour what they would never say in the first.
Ethnographic accompaniment
Sustained presence with older adults and informal caregivers across healthcare and daily-life settings. The method surfaces the gap between what a system records and what a person actually experiences.
Participatory design
Building with, not for. Laguna's interface, language choices, and "do not disturb" posture were shaped by the people who will use it, not by design convention.
Narrative reframing
Training in screenwriting taught me to ask what a situation is actually about, rather than what it appears to be. What looks like non-compliance is often a question about whether a person was ever given the tools to comply.
Technical prototyping
Full-stack development (PHP/MySQL), IoT hardware integration (ESP32), and AI-assisted tools. The technical work is in service of the research question, not the other way around.
I would welcome the chance to discuss how my work might contribute to yours.
ailun@me.com