Skip to main content

About the project

🩺 Med-Triage

AI-Driven Emergency Intake & Intelligent Patient Stratification System


🚨 The Problem

Emergency Departments (EDs) around the globe are facing unprecedented overload:

  • Dangerous Delay Times: Patients frequently wait hours before receiving initial clinical evaluation, risking rapid deterioration.
  • Nurse Burnout: Up to 40% of intake time is swallowed by repetitive administrative documentation and manual triage data entry.
  • Cognitive Fatigue: High patient volumes under chaotic conditions lead to misassigned priority levels and overlooked critical subtle symptoms.

💡 The Solution

Med-Triage re-imagines hospital emergency intake by placing an intelligent, real-time clinical assistant at the front door of the ER:

  • ⚡ Instant Priority Stratification: Converts unstructured patient symptom descriptions and live vitals into accurate, standardized ESI (Emergency Severity Index) priority tiers within seconds.
  • 🧠 Agentic Diagnostic Reasoning: Employs multi-agent LLM pipelines to double-check edge-case red flags (e.g., atypical cardiac presentations, early sepsis indicators).
  • 📊 Live Clinical Command Dashboard: Provides triage personnel with real-time queue visibility, actionable automated risk badges, and one-click intake sign-offs.

✨ Key Features

Feature Description
Multimodal Intake Accepts free-text patient voice/text notes alongside vital metric streams (HR, BP, SpO2, Temp).
Real-time Risk Engine Calculates immediate urgency classifications and flags high-risk anomalies automatically.
Clinical Summary Generator Synthesizes complex intake logs into clear, actionable bulleted handoff notes for attending doctors.
Interactive ER Dashboard Prioritizes patients dynamically based on live vitals updates and changing clinical urgency.
Privacy-First Architecture Built with HIPAA compliance principles in mind—minimizing PII storage and utilizing isolated inference pipelines.

🛠️ Tech Stack

Frontend

  • Framework: React / Next.js (TypeScript)
  • Styling: TailwindCSS & Shadcn UI
  • State & Realtime: WebSockets / React Query

Backend & AI Pipeline

  • Server Framework: FastAPI / Node.js
  • AI Engine: LangChain / LlamaIndex + LLM Orchestration
  • Vector Database: Qdrant / Pinecone (Clinical Knowledge Retrieval)
  • Database: PostgreSQL / Supabase

Built with

geminigemma

AI tools

Coded withGitHub CopilotAntigravityGemini 3.5 Flash LiteGemma 4 26B A4B

Prize Categories Entered

  • Deepmind Google

    Best Use of Gemma 4

    Submitted
  • Event prize

    Best Use of Graphify

    Submitted