Med-triage
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
- Submitted
Best Use of Gemma 4
- Event prizeSubmitted
Best Use of Graphify