MediAgent

Multi-agent medical diagnosis platform that runs specialist AI agents (Cardiology, Psychology, Pulmonology) in parallel and fuses their findings into one comprehensive report.

Table of Contents

• Overview
• Demo / Video
• Tech Stack
• Features
• Project Structure
• Getting Started
• Configuration
• Running the App
• API Routes

Overview

MediAgent combines multiple domain-specific agents behind a Flask API and a React/Tailwind UI. Users upload or paste medical reports; agents analyze concurrently, and a multidisciplinary synthesizer returns a single Markdown-formatted diagnosis.

Demo / Video

• Demo video: https://hostwebs.site/PSXbV4

Tech Stack

• Backend: Flask, LangChain, Google Gemini 2.5 Flash, Flask-CORS, python-dotenv
• Frontend: React 18, React Router DOM, Tailwind CSS, React Markdown, Fetch API
• Tooling: Python venv, npm, dotenv

Features

• Parallel specialist agents + multidisciplinary synthesis
• File upload (.txt) or raw text input; Markdown output
• Responsive UI with loading/error states
• Env-driven configuration for API keys and endpoints

Project Structure

MediAgent/
├── Backend/
│   └── model_and_api_multiagentic_diagnosis/
│       ├── app.py                    # Flask API server
│       ├── requirements.txt          # Python deps
│       ├── Utils/
│       │   └── Agents.py             # Agent classes
│       └── Medical Reports/          # Sample reports
├── Frontend/
│   ├── src/
│   │   ├── pages/
│   │   │   ├── Home.js               # Upload page
│   │   │   └── Results.js            # Results page
│   │   ├── services/
│   │   │   └── api.js                # API calls
│   │   ├── App.js                    # Main app
│   │   └── index.js                  # Entry point
│   ├── package.json                  # Node deps
│   └── tailwind.config.js            # Tailwind config
└── README.md

Getting Started

Clone

git clone https://github.com/<your-org>/MediAgent.git
cd MediAgent

Backend Setup

cd Backend/model_and_api_multiagentic_diagnosis
python -m venv venv
# Windows
venv\Scripts\activate
# macOS/Linux
source venv/bin/activate
pip install -r requirements.txt

Frontend Setup

cd Frontend
npm install

Configuration

Create .env in Backend/model_and_api_multiagentic_diagnosis:

GOOGLE_API_KEY=your_google_api_key_here

(Optional) .env in Frontend:

REACT_APP_API_URL=http://localhost:5000

Running the App

Backend (port 5000):

cd Backend/model_and_api_multiagentic_diagnosis
venv\Scripts\activate  # or source venv/bin/activate
python app.py

Frontend (port 3000):

cd Frontend
npm start

API Routes

• GET / — Health check
  Response: {"status":"API is running","message":"Use POST /process_string (JSON body) or POST /process_file (File Upload) for analysis."}

• POST /process_file — Analyze uploaded .txt report
  Body: multipart/form-data with file
  Response: {"status":"success","diagnosis":"...","filename_processed":"report.txt"}

• POST /process_string — Analyze raw text
  Body: {"report_content":"..."} Response: {"status":"success","diagnosis":"..."}