Civillion Routing
Alternate Routes
Governments can place a custom or default building in any available spot

Inspiration

It starts with a common situation.

Someone feels something is wrong. Chest pain. Trouble breathing. A fever that has lasted too long. They search for the nearest emergency room and drive there, hoping it will help.

But the nearest hospital is often not the best hospital.

Many ERs operate far above capacity. Patients arrive at already overwhelmed hospitals while another ER fifteen minutes away may have open beds. The result is long wait times, overcrowded staff, and delayed treatment.

The problem is not only medical.

It is logistical.

People cannot see which hospitals are overwhelmed, and cities lack tools to understand where new emergency capacity would actually reduce strain.

We asked a simple question:

What if emergency care could be routed like traffic?

That idea became ERoute.

The Solution

ERoute is a spatial decision platform for emergency healthcare.

It helps patients reach the right hospital faster and helps governments design better emergency infrastructure.

The system has two main modes.

Civilian Mode

Civilian Mode helps people decide where they should go for treatment right now.

Location Detection

The system automatically detects the user’s location using GPS.
If GPS is unavailable, users can manually enter a postal code.

This allows the platform to calculate nearby hospitals and driving routes instantly.

Voice Based Triage

Users interact with a conversational voice triage assistant.

Instead of filling out forms, the system asks questions naturally:

symptoms
heart rate
breathing difficulty
pain indicators

Speech is converted to text and analyzed by the AI triage system, while responses are spoken back to the user using natural voice synthesis.

This makes the experience easier during stressful situations.

Symptom Severity Classification

The AI triage engine classifies cases into:

Critical
Urgent
Non-urgent

Each classification includes reasoning explaining why the condition falls into that category.

Severity directly affects hospital routing decisions.

Smart Hospital Routing

ERoute ranks the top three hospitals based on multiple factors:

driving distance
predicted wait time
ER occupancy
hospital specialization

Routing weights shift depending on severity.

Critical patients prioritize shortest travel time, while non-urgent cases prioritize lower congestion.

Route Visualization

Instead of showing a static path, ERoute displays a dynamic route map with traffic-aware segments.

The route updates as the user drives, incorporating live congestion data so the recommendation adapts in real time.

Traffic segments are color-coded to show congestion intensity.

Hospital Traffic Forecast

Users can scrub through a 60-minute congestion forecast in 5-minute intervals.

This allows them to see how ER congestion may change before arriving.

Emergency Actions

For critical situations, ERoute provides a 911 emergency call button.

The interface also displays direct phone numbers for nearby ER departments so patients can contact hospitals immediately if necessary.

Government Mode

Government Mode helps planners understand how hospital infrastructure affects the entire city.

It combines simulation, infrastructure design, and spatial analytics.

Parametric Hospital Designer

Users can design a hospital building directly in the interface.

Parameters include:

building width and length
number of floors
architectural style (modern, classic, industrial)
wall materials and textures
window styles

This allows planners to explore different hospital configurations quickly.

Hospital Resource Configuration

The system also models hospital capacity.

Users can configure:

total beds
ER beds
operating rooms
trauma rooms
ambulances
doctor and nurse counts

Smart Resource Allocation

Hospital resources are interdependent.

The system enforces realistic constraints such as:

beds increase staffing demand
nurses are shared across trauma rooms, operating rooms, and emergency departments
rooms without enough staff are flagged as unusable

This prevents unrealistic infrastructure designs.

Voice Based Hospital Design

Government users can also describe hospitals using natural language.

For example:

“Build a 5-floor modern hospital with 200 beds.”

The AI converts this request into a full building configuration automatically.

Real Time 3D Hospital Visualization

The hospital is rendered in a live 3D viewport using React Three Fiber.

As parameters change, the building updates instantly.

Users can inspect each floor and view the structure spatially, making it easier to understand how the hospital layout works.

This immersive visualization helps planners reason about infrastructure in ways a spreadsheet cannot.

Blueprint Export

Designed hospitals can be exported as GLB 3D models.

These become reusable hospital blueprints that can be placed into the city simulation.

Immersive 3D Emergency Network Simulation

ERoute also provides a 3D city simulation environment that models how patients move across a hospital network.

Hospitals appear on the map with color-coded congestion circles representing occupancy levels:

green → low occupancy
yellow → moderate load
orange → heavy load
red → critical congestion

Coverage Heatmap

A demand heatmap visualizes areas where emergency demand is highest.

The map uses color intensity to show where new hospitals would have the greatest impact.

Blue represents lower demand while red highlights high-pressure zones.

Voronoi Network Simulation

The system models patient distribution using a distance decay diversion model.

When a new hospital is placed:

patients within a 15 km radius may divert
up to 30 percent of demand can shift toward the new ER

The simulation calculates before-and-after occupancy for nearby hospitals.

Flow Visualization

Arrows appear across the map showing how patient flow redistributes.

These flow arcs make it clear which hospitals lose or gain patient demand when new infrastructure is introduced.

ElevenLabs Voice Integration

ERoute integrates ElevenLabs voice generation to power the conversational triage assistant.

All assistant responses are generated using ElevenLabs text-to-speech, allowing the system to guide users through intake questions and explain routing decisions using natural speech.

This creates a more accessible experience during emergencies when users may not want to read large amounts of text.

If the ElevenLabs API is unavailable, the application automatically falls back to browser speech synthesis to ensure the system continues working.

How We Built It

ERoute combines mapping, simulation, AI triage, and immersive visualization.

The system is built using:

Next.js + TypeScript for the web platform
Mapbox GL for the interactive city map
React Three Fiber for real time 3D hospital visualization
MongoDB Atlas for hospital and congestion data
Gemini AI for triage classification
ElevenLabs for voice-based triage responses
Ontario Open Data datasets for hospitals and congestion

The simulation engine generates synthetic patient demand and redistributes it across hospitals to estimate congestion changes.

Challenges We Ran Into

Healthcare congestion data is often inconsistent and delayed. To address this, we built a hybrid system that merges live datasets with fallback congestion estimates.

Another challenge was balancing severity with congestion. Sending every patient to the least busy ER can increase travel time for critical cases, so routing weights must adapt dynamically.

Designing realistic hospital simulations also required modeling how resources like beds and staff interact.