PubMed Helper

A full-stack application for searching and viewing scientific articles from PubMed using natural language queries.

Features

• 🔍 Natural language search for PubMed articles
• 🤖 AI-powered brief summaries (using OpenAI GPT-3.5)
• 📄 Full article abstracts
• 📑 Section identification in articles
• 🎨 Modern, responsive UI

Tech Stack

• Backend: NestJS (Node.js)
• Frontend: Vue 3 with Vite
• API: PubMed E-utilities API
• AI: OpenAI GPT-3.5 Turbo (optional, with fallback)

Project Structure

pubmed-helper/
├── backend/          # NestJS backend
│   ├── src/
│   │   ├── ai/        # AI service for generating summaries
│   │   ├── pubmed/    # PubMed API service
│   │   ├── search/    # Search controller and service
│   │   └── main.ts    # Application entry point
│   └── package.json
│
└── frontend/         # Vue 3 frontend
    ├── src/
    │   ├── components/  # Vue components
    │   ├── services/    # API service
    │   └── App.vue      # Main app component
    └── package.json

Setup Instructions

Backend Setup

1. Navigate to the backend directory:

cd backend

2. Install dependencies:

npm install

3. (Optional) Set up OpenAI API key for AI summaries:

   • Get your API key from OpenAI Platform
   • Set it as an environment variable:

   export OPENAI_API_KEY=your_api_key_here

   • Or create a .env file in the backend directory:

   OPENAI_API_KEY=your_api_key_here
   

   • Note: If no API key is provided, the app will use intelligent fallback summaries based on the abstract.

4. Start the development server:

npm run start:dev

The backend will run on http://localhost:3000

Frontend Setup

1. Navigate to the frontend directory:

cd frontend

2. Install dependencies:

npm install

3. Start the development server:

npm run dev

The frontend will run on http://localhost:5173

Usage

1. Start both the backend and frontend servers
2. Open your browser to http://localhost:5173
3. Enter a natural language query (e.g., "diabetes treatment", "cancer research")
4. View the AI-generated brief summaries, full abstracts, and article sections

API Endpoints

GET /search

Search for articles using natural language.

Query Parameters:

• q (required): Search query
• limit (optional): Maximum number of results (default: 5, max: 20)

Example:

GET http://localhost:3000/search?q=diabetes%20treatment&limit=5

License

MIT