The Future of
Autonomous Triage
Addressing the challenge: "How might we design user-centric solutions that improve the quality of life in Singapore?"
The Problem
The "3-Hour" Bottleneck
As Singapore ages, our polyclinics are overwhelmed. The average wait time for a consultation often exceeds 3 hours.
Currently, highly trained nurses spend thousands of hours manually asking:"What are your symptoms?" and "Do you have any drug allergies?".
Doctors often lack immediate context on a patient's history, relying on verbal confirmation which is slow, unreliable, and prone to memory error.
Why TriageX is Better
Technical Execution
Voice Input
ElevenLabs Scribe v1
Consensus Engine
Gemini 2.5 + GPT-OSS + Llama 4
Triage Ticket
P-Score & Notes
Safety-First Consensus Algorithm
We poll three state-of-the-art models via Vercel AI SDK. We prioritize safety over averages.
THEN final_score = "P1" (Override)
ELSE final_score = CEIL(AVG(votes))
Structured Output Enforcement
Unlike standard chatbots, TriageX is the only solution using Zod schema validation to enforce strict JSON outputs. This ensures data is 100% compatible with hospital databases.
"acuity": "P2",
"actions": ["RICE", "X-Ray"]
}
- Next.js & TypeScript
- Tailwind CSS
- Framer Motion
- ElevenLabs Scribe v1
- Eleven Multilingual v2
- Audio Isolation Engine
- HAPI FHIR Server
- HL7 Interoperability
- Docker Container
Kiosk User Guide
Select Profile
Simulate scanning an NRIC. The system pulls relevant history (e.g., Diabetes) from the local FHIR database.
Click-to-Speak
Tap the microphone once to start describing symptoms. Tap again to stop. No need to hold the button.
Receive Ticket
Review the AI's consensus log and "Print" your Triage Slip for the doctor.
FHIR Simulation
We run a local Docker instance of hapiproject/hapi to simulate the National Electronic Health Record (NEHR).
> 200 OK
> Conditions: [Hypertension, Asthma]