Simulation · not the live system · resets restore demo data for everyone ← Live stations

AAIS

Ghana Pilot: Emergency Triage & Routing

Investor Briefing: Autonomous Triage, Deterministic Routing & Hospital Coordination

How AAIS Works

AAIS is a nationwide emergency workflow engine designed to eliminate transit delay, ensure clinical-bed fit, and notify emergency rooms before ambulances arrive.

This simulator demonstrates a live incident moving through our 7 architectural stages, highlighting how AI-driven clinical reasoning integrates with deterministic routing rules.

Core: FastAPI, Python 3.11

AI Triage: Configurable LLM provider (NVIDIA NIM or DeepInfra)

Routing: Dynamic Multi-Factor Scoring

7-Stage Architectural Pipeline

1

NAS Dispatch Links Patient ID (Ghana Card, NHIS) Mock Database API Integration

2

Ambulance Scene Captures vitals, chief complaints Telemetry Seed Paramedic Mobile App

3

AI Triage Extracts acuity, pathways, capability requirements LLM Provider FHIR Triage Pipeline

4

Deterministic Routing Weighs capacity, clinical fit, ETA, ER load Scoring Engine Live GIS Routing

5

Hospital Pre-Alert Auto-generates & sends priority alerts to ER WebSocket Gateway WhatsApp / EHR API

6

Ambulance Transport Tracks patient vitals and ETA in real-time Step Telemetry Active GPS Streams

7

Digital Handover SBAR document generation & clinical signoff AI Summary HL7 FHIR EHR Write

AAIS

AAIS Command Center

National Emergency Routing Pilot — Ghana Dispatch

🗺️ Visual Flow 🚑 Paramedic CAD 🎮 Bed Simulation 📋 Bed-Board 📲 QR Scan 📺 Inbound TV

LLM Provider Checking

Resolving active model...

STEP 0/7 Select an Incident

Auto-Play

Demo Speed:

2.0s

🎙️

Live Flow Explanation

Choose a patient incident scenario from the left panel, then initialize the workflow to launch the investor walkthrough.

💡

Why It Matters (Clinical & Business Value)

Understanding the real-world impact helps prove the commercial viability of nationwide health-routing platforms. Select a scenario to begin.

📞

NAS Dispatch

Waiting

🚑

Ambulance Scene

Pending

🧠

AI Triage

Pending

⚖️

Deterministic Routing

Pending

🔔

Hospital Pre-Alert

Pending

🗺️

Transport Track

Pending

📝

EHR Handover

Pending

Active Incidents 0

Ambulances Online 0

Open ER Beds 0

Average ER Load 0%

Live Operational Picture

Real-time GPS tracking centered on incident node

Hospital Capacity Dashboard

Regional clinical resource beds

NAS Ambulance Fleet

Vehicle coordinates & status

Patient Clinical Profile

Awaiting incident initiation

No incident selected.

Deterministic Hospital Recommendation Engine

Weighing clinical capability match, travel times, and load filters

Routing Engine Formula & Weights

AAIS first limits routing candidates to hospitals in the incident region, then maps medical indicators to hospital capabilities and filters out hospitals lacking absolute clinical requirements (e.g. Stroke clinical pathway requires CT scanner). Regional survivors are graded via:

Weight Metric	Target Value	Impact
Clinical Capability (40%)	Matches triage pathways + staff	Ensures specialized center assignment
Capacity Resources (25%)	Available ER beds + ventilators	Guarantees resources are free upon arrival
Ambulance Travel ETA (20%)	GIS calculated drive ETA	Minimizes travel delays
ER Load Balance (10%)	Busy ratio of the Emergency Room	Prevents overcrowding and ambulance queueing
Record Continuity (5%)	EHR preferred hospital history	Leverages preexisting clinician relationships

Hospital Pre-Alert & SBAR Handover Note

AI-triage generated notifications and clinical transitions

PRE-ALERT MESSAGE

NOT SENT

Pre-alert message will automatically be drafted and sent to the destination emergency room during Step 5 (Pre-Alert).

EHR DIGITAL HANDOVER (SBAR)

AWAITING HANDOVER

Structured clinical SBAR (Situation, Background, Assessment, Recommendation) note will compile upon ambulance arrival during Step 7 (Handover).

Command Audit Timeline

Traceable event logs of pilot system

National Scale Production Architecture

Evolving from the AAIS Pilot Simulator to a highly available, nationwide emergency infrastructure.

🗄️

1. Data Persistence Layer

Pilot: In-Memory / Python Session Store → Production: PostgreSQL + PostGIS (GIS Spatial) + Redis Cache

Moving from ephemeral local session dicts to structured database schemas with indexing on latitude/longitude spatial coordinates, optimized for complex municipal geometries.

🚀

2. Event Streaming & Telemetry

Pilot: Periodic HTTP API polling → Production: Apache Kafka Cluster / WebSockets

Supporting real-time GPS telemetry from thousands of mobile NAS units and hospital bed sensors. Event streams ensure instantaneous UI synchronization without taxing server loads.

🔒

3. Enterprise Security & DPA Compliance

Pilot: Public routes, mock IDs → Production: OpenID Connect, AES-256 Patient Cryptography

Full integration with Keycloak/Auth0 for Role-Based Access Controls. Encryption-at-rest for sensitive clinical patient records matching the strict rules of the Ghana Data Protection Act.

🔌

4. Integrated National APIs

Pilot: Mocked JSON Seeding → Production: GH Card Registry API + NHIS Eligibility Gateway

Direct connectivity with the National Identification Authority (NIA) to pull verified demographic snapshots via fingerprints or card numbers, linking patients automatically with national insurance eligibility profiles.

🤖

5. Clinical LLM Guardianship

Pilot: NVIDIA NIM or DeepInfra provider → Production: Redundant LLM Ensemble + Human Verification

Triage prompts route to parallel specialized clinical models with fallback strategies. High-acuity decisions require physical dispatcher confirmation on confidence drop, avoiding model hallucination risks.

📟

6. ER ERX pre-alert delivery

Pilot: Web-socket telemetry rendering → Production: WhatsApp ER Dispatch Bot, SMS & EHR FHIR Push

Ensuring ER clinicians get alerts immediately by broadcasting to critical clinical WhatsApp groups, Twilio SMS fallback channels, and writing directly to their internal electronic hospital health record systems.