CarePilot Urgent is a FastAPI + React urgent care workflow app:

• Patient portal – intake, QR/token for check-in (no kiosk link; kiosk is hospital-only)
• Kiosk – QR scan + manual code entry, voice assistant, vitals (hospital-only, e.g. Jetson Nano)
• Waiting room display – token-only queue (privacy-safe)
• Staff – queue, status updates, analytics (password-protected)

Sharing on GitHub: See LICENSE (all rights reserved) and SHARING.md for keeping secrets out of the repo and your rights when the repo is public.

python -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate
pip install -r requirements.txt
cp .env.example .env
# Edit .env: STAFF_ACCESS_PASSWORD, GEMINI_API_KEY (optional)
uvicorn app:app --reload --host 0.0.0.0 --port 8000

Open http://localhost:8000/

cd frontend
npm install
npm run build
cd ..
uvicorn app:app --reload --host 0.0.0.0 --port 8000

When frontend/dist exists, the server serves the React SPA (Tailwind, shadcn/ui, Framer Motion). Same URLs; APIs unchanged.

Copy .env.example to .env and set as needed:

cp .env.example .env

In production, if STAFF_ACCESS_PASSWORD is not set (or left default), the app uses a built-in fallback password when unset; set STAFF_ACCESS_PASSWORD in your environment; optionally set STAFF_FALLBACK_PASSWORD for a fallback.

• Local (development): use the password in .env (e.g. STAFF_ACCESS_PASSWORD=1234 or your choice).
• Production (e.g. Render): if you have not set STAFF_ACCESS_PASSWORD in the dashboard, set STAFF_ACCESS_PASSWORD in the host environment; optionally set STAFF_FALLBACK_PASSWORD for a fallback when unset.

Staff URL: /staff/login → after login you get /staff (queue) and /analytics.

Kiosk is not linked from the patient portal. Patients are told to bring their token/QR to the check-in kiosk at the hospital. Open /kiosk-station only on the kiosk device (e.g. Jetson Nano).

Use one base URL (e.g. https://carepilot-urgent.onrender.com or http://localhost:8000):

Flow: Patient does intake on phone → gets token/QR → at kiosk (Computer 2) scans or enters token → staff sees them on Computer 1 → waiting room screen shows queue.

Build:

docker build -t carepilot-urgent:latest .

Run (interactive):

docker run -p 8000:8000 --env-file .env carepilot-urgent:latest

Run with persistent database:

docker run -p 8000:8000 --env-file .env -v carepilot-data:/data carepilot-urgent:latest

Run in background:

docker run -d -p 8000:8000 --env-file .env --name carepilot carepilot-urgent:latest
# Stop: docker stop carepilot

Using Docker Compose:

docker compose up --build -d

Then open http://localhost:8000. The image builds the React frontend and runs FastAPI; it expects PORT=8000 (or set PORT in env).

See DEPLOY.md for:

• Railway: connect repo, set STAFF_ACCESS_PASSWORD, APP_SECRET_KEY, optional volume for DB_PATH=/data/carepilot.db
• Render: Docker runtime, env vars, optional disk at /data for SQLite

After deploy, use the same paths: /, /intake, /kiosk-station, /display, /staff/login, etc.

To send vitals from hardware (SpO2, HR, temp, BP) into the app:

• API: POST /api/vitals/submit/json with JSON body: token, device_id, spo2, hr, temp_c, bp_sys, bp_dia, etc.
• Script: Run the sensor bridge on the Nano (or any machine with sensors). See SENSORS.md for:
  • CAREPILOT_URL, CAREPILOT_TOKEN, CAREPILOT_INTERVAL, CAREPILOT_DEVICE_ID
  • Simulated vs real sensors (Max30102, DS18B20, etc.)
  • Example: export CAREPILOT_TOKEN=UC-1234 && python scripts/sensor_bridge.py

Using your MacBook camera (local development):

• Defaults are already set: CAMERA_INDEX=0, CAMERA_PIPELINE empty. The app uses the built-in webcam and, on macOS, tries the AVFoundation backend for better compatibility.
• macOS: Grant Camera access to Terminal (or Python) in System Settings → Privacy & Security → Camera. If the camera fails to open, the error message will remind you.
• Open http://localhost:8000/kiosk-station — the live view and QR scan use your MacBook camera.
• If the camera still doesn’t open (e.g. with opencv-python-headless), install the full OpenCV for local dev: pip install opencv-python (then run the app again). The Docker/deploy build can keep using headless.

Jetson (Nano / Orin Nano) production kiosk:

• Camera: ls /dev/video* and v4l2-ctl --list-devices. Optional CAMERA_PIPELINE for GStreamer (CSI cameras). See docs/JETSON-CAMERA.md for connecting the Jetson camera so anyone opening the kiosk can see it (run app on Jetson, or remote-stream options).
• OpenCV on Jetson: sudo apt-get install python3-opencv (with GStreamer if using CSI).
• Kiosk: Open http://<jetson-ip>:8000/kiosk-station. Camera + code entry + voice run in the browser.
• Sensors: Use scripts/sensor_bridge.py and SENSORS.md to send vitals by token.

• Data is stored in SQLite (DB_PATH). Use a single app instance for in-memory websocket and camera coordination.
• For production at scale, use a managed database and shared pub/sub instead of in-memory state.

• Repo: https://github.com/Yagbe/CarePilot-Urgent (or your fork)
• Clone → create branch → make changes → push → open a Pull Request to main.