Skip to content

Pooja-Vachhad/MedBridge

Repository files navigation

MedBridge

Transforming medical emergencies from chaos to clarity


Overview

MedBridge is an AI-powered emergency healthcare platform designed to remove uncertainty during medical emergencies. In under 60 seconds, it connects patients to the right hospital, coordinates ambulance dispatch, and prepares medical teams before arrival—ensuring treatment starts faster when every second counts.


The Problem

When my father had an accident, we faced:

  • Missing equipment at the nearest hospital
  • 30-45 minutes lost switching hospitals and handling paperwork
  • Endless questions about specialists, bed availability, and insurance
  • Confusion and panic during the critical Golden Hour

In emergencies, families don't need the nearest hospital—they need the right hospital.


How MedBridge Works

Step 1: Describe the Emergency

symptoms_One.mp4

Families describe symptoms in plain language—no forms, no medical jargon.

Behind the scenes:

  • Powered by Gemini 2.5 Flash (Vertex AI)
  • Smart Triage Layer analyzes symptoms in ~2 seconds
  • Determines urgency level:
    • Emergency — chest pain, severe bleeding, unconsciousness
    • Urgent — high fever, persistent vomiting, breathing difficulty
    • Non-Emergency — mild symptoms, common illness

Result: Clear, actionable decision with no confusion.


Step 2: Find the Right Hospital with RAG

untitled-video-made-with-clipchamp-1_WyEqwhDE.mp4

The nearest hospital isn't always the right one.

MedBridge uses Retrieval-Augmented Generation (RAG) to recommend hospitals based on:

  • Specialty availability (Cardiology, Neurology, Trauma, etc.)
  • Emergency bed availability (real-time status)
  • Insurance acceptance (Ayushman, CGHS, ESI, private, cashless)
  • Distance and infrastructure (ICU, ventilators, cath lab)

How it works:

  1. Symptom → Specialty Mapping: Gemini identifies required care (e.g., chest pain → Cardiology + Emergency Medicine)
  2. Smart Retrieval & Filtering: Hospitals filtered by distance (2km–15km), insurance, hospital type, and minimum rating
  3. Capability Check: Only hospitals with:
    • Open emergency beds
    • On-call specialists
    • Required equipment
  4. Intelligent Scoring: Results ranked by urgency, proximity, and capability

Each recommendation explains itself:

Trauma center • 2.5 km away • Cardiology available • CGHS accepted

Why this matters: Pure LLMs can invent hospital details. MedBridge doesn't. Gemini reasons only over retrieved, verified data—no hallucinations, only grounded results.


Step 3: One-Click Ambulance Dispatch

untitled-video-made-with-clipchamp-2_RkBe7gVx.mp4

After selecting a hospital, families click "Call Ambulance"—and everything happens instantly.

What happens behind the scenes:

  • Ambulance dispatched from selected hospital
  • Bed reserved (e.g., "Floor 3, Bed B-205")
  • Doctor assigned (e.g., "Dr. Sharma waiting in Emergency Wing")
  • Patient medical profile auto-shared with hospital
  • Zero paperwork needed upon arrival

Family receives:

  • Vehicle number and driver contact
  • Live GPS tracking with ETA countdown
  • Complete hospital admission details

The breakthrough: When the family arrives at the hospital, treatment starts immediately—no wasting golden minutes on forms.


Step 4: AI Guardian - Voice Companion During Transit

ai-sathi_Zhsb7k9O.mp4

During the ambulance journey, panic sets in. Questions flood the mind.

AI Guardian is a real-time voice assistant powered by:

  • ElevenLabs TTS (natural, calming voice)
  • Gemini Context Engine (intelligent, context-aware responses)

What AI Guardian provides:

  • Emotional support with an empathetic, calming tone
  • Answers to medical and logistical questions
  • Real-time guidance until treatment begins

Common questions answered:

  • "Will they accept my insurance?"
  • "What documents do I need?"
  • "How much will this cost?"
  • "Is my loved one going to be okay?"

Step 5: Building Community Care

Image

MedBridge isn't just for your own emergencies—it's designed to encourage helping strangers.

The Story: When my father had his accident, two kind strangers helped but had to leave urgently. We couldn't thank them properly. Across India, bystanders hesitate to help accident victims due to legal concerns or feeling no responsibility.

How Health Points Work:

  1. Someone uses MedBridge to create an emergency for a stranger (not family/self)
  2. System automatically awards 50 Health Points
  3. Points stored in user's HealthPoints account
  4. Redeem points for:
    • Free health checkups at partner clinics
    • Diagnostic tests (blood work, X-rays, ECG)
    • Priority appointment booking (emergency or routine)
    • Ambulance credits for future emergencies

The Goal: Transform hesitant bystanders into confident helpers. Build community safety nets where helping others is rewarded and celebrated.


Step 6: Full Observability with Datadog

datadog-feature_MVTks4Q4.mp4

Behind every successful emergency response, there's a complex AI system that needs to perform flawlessly.

Image

MedBridge uses Datadog for end-to-end LLM observability—ensuring our AI never fails when lives depend on it.

Image

What we monitor:

  • LLM response latency
  • Error rates
  • User flow completion
  • System health signals

Result: Failures are detected early and reliability is maintained.

Image

Technology Stack

Component Technology Purpose
Backend Django (Python) Core application logic, REST APIs
Frontend HTML, CSS, JavaScript Responsive user interface
AI Analysis Google Vertex AI (Gemini 2.5 Flash) Symptom triage, hospital recommendations
Voice AI ElevenLabs (eleven_turbo_v2_5) Text-to-speech for AI Guardian
Database Django ORM + SQLite (PostgreSQL-ready) Patient records, hospital data, health points
Observability Datadog Metrics API v2 LLM monitoring, detection rules, alerts
RAG System Custom Python + JSON Knowledge Base Hospital-symptom matching with real-time scoring
Deployment Google Cloud Platform Compute Engine, Cloud Storage
Authentication Django Session Management Secure user authentication

Future Vision

  • Health Dashboard: Continuous vitals monitoring + early warnings
  • Doctor Chat/Video: Remote consultations, appointments
  • Multilingual: Hindi, Tamil, Telugu support
  • Live Tracking: Real-time ambulance GPS with ETA
  • Hospital APIs: Direct EHR integration

Why MedBridge is Different

MedBridge is not a chatbot. It is an AI-orchestrated emergency coordination system.

In under 60 seconds, it:

  • Understands emergency symptoms (Gemini on Vertex AI)
  • Finds the right hospital using RAG (not hallucinations)
  • Prepares ambulance + hospital before arrival
  • Guides families with a real-time voice AI (ElevenLabs)
  • Monitors reliability with Datadog observability

Outcome: Treatment starts faster. No chaos. No paperwork. No guessing.


Project Status

This project is a demonstration prototype using simulated hospital availability and ambulance workflows to showcase the system design.


Mission

"In emergencies, every second counts. MedBridge makes every second count."

About

MedBridge is an AI-powered emergency healthcare platform designed to remove uncertainty during medical emergencies.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors