← Back to Work
AI Systems · 2026Working prototype — full-stack build

KelinaCare — AI Healthcare Operations Platform

A healthcare operations prototype built with Next.js, FastAPI and Supabase. Patients can register, find departments and book appointments; staff use role-based dashboards, with AI support for intake logistics.

A healthcare worker in a clinical setting using a device — representing KelinaCare's patient-facing healthcare web application.Context photograph

The challenge

Registration and appointments depended on manual, phone-driven processes. The system needed a consistent patient identity and an assistant that would not drift into clinical advice.

Objective

Give each patient one record and hospital number, prevent double bookings, and let staff manage appointments through role-based dashboards.

The solution

The application generates patient identifiers in the database and prevents appointment collisions with a unique index. Mutations are recorded in an audit log.

An OpenAI function-calling assistant connects six typed tools for registration, lookup, availability, booking, departments and contacts.

  • The assistant handles logistics, with no diagnosis, prescriptions or test interpretation.
  • Emergency queries route to verified contact information.
  • Staff authentication and doctor/nurse dashboards support the operational workflow.

Key features

  • Patient intake & identityMulti-step registration that creates one record per patient with an auto-generated, unique hospital number — no duplicate identities.
  • Patient lookupSearch and retrieve patient records by hospital number or name so conversations are always tied to the right person.
  • Appointment workflowChoose department, date and time slot with live availability and a database-level guard against double bookings.
  • Departments & contactsBranch-filtered department directory with verified emergency contacts from the hospital's official sources.
  • Role-based staff dashboardsDoctor and nurse dashboards with appointment views and status tracking; auth with register, login, forgot-password and reset flows.
  • AI intake assistantConversational registration, booking, availability and contact lookup via typed tool calls — with strict no-clinical-advice safety rules and emergency escalation.

How the system works

  1. 01 — PatientWeb or chat entry point
  2. 02 — Intake / RegistrationStructured form or AI conversation
  3. 03 — Patient recordUnique KEL-YYYY-NNNNNN number
  4. 04 — Lookup & identifySearch by name or hospital number
  5. 05 — AppointmentDept, date, slot — double-booking blocked
  6. 06 — Care coordinationStaff dashboards + audit trail

Current status & progress

The patient workflow, staff authentication, doctor/nurse dashboards and AI intake assistant are implemented as a working prototype. Admin/receptionist dashboards, automated tests and applied row-level security policies remain pending.

Delivery status

  • CompletePatient registration & records
  • CompletePatient lookup
  • CompleteDepartments & emergency contacts
  • CompleteAppointment booking + availability
  • CompleteStaff auth (register, login, password reset)
  • CompleteDoctor & nurse role dashboards
  • CompleteAI intake assistant (function calling)
  • CompleteAudit logging on records
  • PendingAdmin & receptionist dashboards
  • PendingAutomated test suite
  • PendingRow-level security policies applied