A smart healthcare emergency routing platform for Ontario. ERoute helps both government planners optimize ER placement and patients find the best hospital based on their medical condition, live traffic, and real-time ER congestion.

• Location detection via browser geolocation or postal code geocoding (Mapbox)
• Vitals capture — enter heart rate, respiratory rate, and stress index
• Symptom intake — select from structured symptom cards (chest pain, shortness of breath, fever, dizziness, free text)
• AI-powered triage — GPT-4o-mini classifies severity as critical / urgent / non-urgent
• Multi-factor smart routing that scores every hospital using:
  • Live driving time via Mapbox Directions API (driving-traffic profile — accounts for real-time traffic, road closures, construction)
  • Time-of-day traffic patterns — rush hour penalties (7-9am, 4-6pm weekdays)
  • ER wait time adjustments — projects wait based on hourly ER volume patterns
  • Hospital occupancy — penalizes overcrowded ERs
  • Specialty matching — routes cardiac patients to cardiac centers, trauma patients to trauma centers, etc.
  • Severity-based weight profiles — critical patients prioritize drive time; non-urgent patients prioritize lowest total wait
• Top 3 recommendations with driving time, wait time, total estimated time, and reasoning
• Route visualization — actual driving route drawn on the map (green for recommended, dashed gray for alternatives)
• User location marker — pulsing blue dot shows your position
• 911 button for critical severity cases

• Interactive map — click to place a proposed new ER location
• Capacity slider — set bed count for the proposed facility
• Voronoi simulation — recalculates patient distribution across the hospital network
• Before/after analysis — shows occupancy impact on every existing hospital
• Flow arcs — visualizes patient redistribution on the map

• Mapbox GL with dark theme and 3D building extrusions
• Hospital congestion circles — color-coded by occupancy (green → yellow → orange → red)
• Live data from Ontario Open Data API (with mock fallback)

• Create and edit 3D building models with React Three Fiber
• Export to GLB format for map integration

• Framework: Next.js 16 (App Router)
• Language: TypeScript
• Styling: Tailwind CSS 4
• Maps: Mapbox GL
• AI: OpenAI GPT-4o-mini (triage classification)
• Database: MongoDB (hospitals, congestion snapshots)
• 3D: React Three Fiber + Three.js
• Animations: Framer Motion
• GIS: Turf.js, Mapbox Directions API

• Node.js 18+
• MongoDB instance (local or Atlas)

Create .env.local at the project root:

MONGODB_URI=mongodb+srv://...
OPENAI_API_KEY=sk-...
NEXT_PUBLIC_MAPBOX_TOKEN=pk.eyJ...
MAPBOX_SECRET_TOKEN=pk.eyJ...

npm install
npm run dev

After starting the dev server, seed hospitals and congestion data:

curl -X POST http://localhost:3000/api/clearpath/seed

The smart routing algorithm scores each hospital using a weighted formula:

score = w_drive * drivingTime + w_wait * adjustedWaitTime + w_occ * occupancyPenalty + w_spec * specialtyMismatch

Weights by severity:

• Critical: Get to the nearest ER as fast as possible
• Urgent: Balance drive time with hospital capacity
• Non-urgent: Minimize total time (drive + wait), prefer less crowded ERs