The Human Layer
Real-time accompaniment by trained companions. Appointment preparation, in-visit support, and post-visit processing, with a summary delivered to the family.
Try the serviceDeployed Care Platform · HCI × Digital Ageing · 2024–2026
A deployed care-communication platform, built from scratch. Live in pilot since July 2026.
Two years accompanying aging parents through Taiwan's healthcare system taught me what non-compliance actually looks like from the inside, and what it would take to design around it.
Ai-Lun Huang · Designer × Researcher
ailunhuang.com
Two layers, one service
Real-time accompaniment by trained companions. Appointment preparation, in-visit support, and post-visit processing, with a summary delivered to the family.
Try the serviceA full-stack system built from scratch:
PHP/MySQL backend · ESP32 hardware integration · AI tools for note organisation
Case study frame
A browser-based PWA prototype that helps patients and caregivers prepare a three-point medical briefing before a consultation, use low-attention prompts during the visit, and wrap up confusion or instructions afterward.
Research framing, interaction concept, interface structure, multilingual scenario design, prototype logic, documentation, front-end prototype implementation, and testing.
This prototype tests one branch of a broader research question: how interactive technologies can support dignity, self-representation, and shared understanding when formal access exists but cognitive, linguistic, or emotional pressure makes participation difficult.
This is not a clinical device, diagnostic tool, or triage system. It is a research prototype for exploring how patients and caregivers can prepare, state, and preserve what matters before attention, memory, or time pressure collapses the conversation.
Three systems, one service chain
Communication layer
Patient-stated summaries, 7-day logs, low-attention appointment prompts, and wrap-up.
Operations layer
Bookings, dispatch, service state, post-visit notes, fees, and CSV export.
Field event layer
Departed, arrived, need-help, completed, and GPS fix/no-fix events.
Role journeys
Live artifacts
A login-protected PHP/MySQL CRM is already deployed as the private operations backend and interview walkthrough. It is not the public site's primary entry point. ESP32 GPS events enter that backend's service timeline.
The next question isn't how to build this. It's how an adaptive, human-centred AI might adjust its support as a person's motivation, capability, and life circumstances change across months and years. That is the PhD question.
Trust boundaries
The IoT API is not a future ornament. It is now deployed inside the private Companion CRM: ESP32 can POST WiFi events to /api/companion/iot_events.php, and the backend already stores heartbeat, GPS fix/no-fix, satellites, HDOP, need_help, and service-state events. The public page explains the business model, booking flow, and service boundaries.
Laguna presents patient-stated information; it does not diagnose, triage, score risk, or advise medication.
Companion CRM is now a private login-protected PHP/MySQL operations backend for service requests, professional caregivers, summaries, and IoT events.
ESP32 GPS adds event context; it is not continuous tracking or a wandering-prevention guarantee.
No-fix GPS events are still accepted so service reliability does not depend on unstable satellite reception.