Skip to content

Healthcare · Self-service kiosk

Hospital Queue Kiosk (QMS)

Self-service kiosk for patient registration, doctor selection, and queue check-in at Medistra Hospital, used without staff assistance.

Role
Android Developer
Organisation
PT Sentuh Digital Teknologi
Platforms
Android
  • 11,000+patient visits per monthMonthly average of patient visits processed through the kiosk at Medistra Hospital.

The deployment described here is the self-service kiosk at Medistra Hospital. Other clients of the same platform are withheld under NDA.

Context

A frozen kiosk or a stale queue number turns straight into a line of frustrated people, many of them elderly or unwell. The kiosk had to:

  • Show the correct state in real time, on every screen.
  • Keep running through unreliable on-site networks.
  • Be usable by elderly and unwell patients with no training and no staff help.

Contribution

I built the Android kiosk app: the patient flow, the real-time connection to the backend, and the app architecture.

Technical approach

  • Returning patientFound by medical record number, NIK, or name and date of birth
  • New patientRegisters first
  1. Specialty
  2. Doctor schedule
  3. Time slot
  4. Payment method
  5. Queue registration and printed ticket
Two ways in, one linear path out: large touch targets, few steps and no dead ends. One shared session carries the patient through the flow and is cleared at the end; live state comes over WebSocket.
  • One linear flow per patient. Large touch targets, few steps, and no dead ends, from identification to a printed queue ticket. The flow was designed for elderly users and patients who are not feeling well.
  • Two patient journeys. Returning patients are found by medical record number, national ID number (NIK), or name and date of birth. New patients register first. Both continue through specialty, doctor schedule, time slot, and payment method before the queue registration.
  • Real-time state over WebSocket. A persistent WebSocket connection keeps kiosks, displays, and the backend on one live source of truth. Reconnection with backoff and state re-hydration means a dropped connection does not leave a screen showing stale data.
  • Explicit UI states with MVI. Each screen exposes one sealed UI state through a StateFlow, so loading, success, and error are distinct, testable states instead of implicit side effects.
  • Clean Architecture with module boundaries. Business logic lives in a domain module without Android dependencies, data access in a data module, and the UI consumes use cases through Hilt.
  • One activity, one shared session. A single activity with Jetpack Navigation shares one ViewModel across the flow, so patient context stays consistent and is cleared when the session ends.
  • No rebuild to switch environments. An OkHttp interceptor reads the base URL from the kiosk settings screen.

Outcome

The kiosk runs in production at Medistra Hospital and handles a monthly average of 11,000+ patient visits for registration and queue check-in, without staff assistance.