ClinicaQuery-AI

An AI-powered medical research assistant that lets you upload PDF papers and ask questions about them using Retrieval-Augmented Generation (RAG). Built with Next.js, FastAPI, ChromaDB, and Google Gemini.

---

Features

• PDF Upload & Indexing — drag-and-drop PDFs and index them into a vector database in seconds
• RAG-powered Q&A — ask natural language questions and get answers grounded in the uploaded papers
• Source Attribution — every answer shows exactly which passages it drew from
• Confidence Scoring — guardrail system rates how well each answer is supported by the sources
• Google OAuth — secure sign-in via Google, powered by NextAuth.js (free, no billing required)
• Markdown Rendering — AI answers render with bold, lists, and headers for readability
• Dark Sidebar Layout — modern indigo/slate UI with navigation, user avatar, and sign-out

---

Tech Stack

Frontend

Tech	Purpose
Next.js 16 (App Router)	React framework
TypeScript	Type safety
Tailwind CSS v4	Styling
NextAuth.js (Auth.js v5)	Google OAuth
Axios	API calls
Lucide React	Icons

Backend

Tech	Purpose
FastAPI	REST API
ChromaDB	Vector store
Google Gemini AI	LLM + embeddings
LangChain	PDF chunking & RAG pipeline
Python-dotenv	Environment config

---

Project Structure

ClinicaQuery-AI/
├── backend/
│   ├── core/
│   │   ├── ingestion.py      # PDF chunking + embedding
│   │   ├── generator.py      # Gemini answer generation
│   │   └── guardrails.py     # Confidence scoring
│   ├── routes/
│   │   ├── upload.py         # POST /api/upload
│   │   └── query.py          # POST /api/query
│   ├── vectorstore/
│   │   └── chroma_store.py   # ChromaDB wrapper
│   └── main.py               # FastAPI app entry point
│
└── frontend/
    ├── app/
    │   ├── api/auth/[...nextauth]/
    │   │   └── route.ts      # NextAuth handler
    │   ├── components/
    │   │   ├── AppShell.tsx  # Auth guard + sidebar layout
    │   │   ├── Sidebar.tsx   # Dark navigation sidebar
    │   │   ├── ChatWindow.tsx# Main chat interface
    │   │   ├── SourceCard.tsx# Collapsible source snippets
    │   │   └── ConfidenceBadge.tsx
    │   ├── login/
    │   │   └── page.tsx      # Google sign-in page
    │   ├── upload/
    │   │   └── page.tsx      # Drag-and-drop PDF upload
    │   └── layout.tsx
    ├── auth.ts               # NextAuth config
    └── proxy.ts              # Route protection

---

Getting Started

Prerequisites

• Node.js 18+
• Python 3.10+
• A Google Cloud project (free tier) — for OAuth
• A Google Gemini API key — get one free

---

1. Clone the repository

git clone https://github.com/Sowaiba-01/ClinicaQuery-AI.git
cd ClinicaQuery-AI

---

2. Backend setup

cd backend
python -m venv .venv
.venv\Scripts\activate        # Windows
# source .venv/bin/activate   # macOS/Linux

pip install -r requirements.txt

Create backend/.env:

GEMINI_API_KEY=your-gemini-api-key-here

Start the backend:

uvicorn main:app --reload --port 8000

Backend runs at http://localhost:8000

---

3. Frontend setup

cd frontend
npm install

Create frontend/.env.local (copy from .env.local.example):

AUTH_SECRET=your-random-secret        # openssl rand -base64 32
AUTH_GOOGLE_ID=your-client-id.apps.googleusercontent.com
AUTH_GOOGLE_SECRET=your-client-secret
AUTH_URL=http://localhost:3000

Setting up Google OAuth (free)

1. Go to console.cloud.google.com and create a project
2. APIs & Services → OAuth consent screen → External → fill in app name & email → Save
3. APIs & Services → Credentials → Create Credentials → OAuth 2.0 Client ID
   • Application type: Web application
   • Authorised redirect URI: http://localhost:3000/api/auth/callback/google
4. Copy the Client ID and Client Secret into .env.local
5. OAuth consent screen → Test users → Add users → add your Gmail

Start the frontend:

npm run dev

Frontend runs at http://localhost:3000

---

4. Open the app

Navigate to http://localhost:3000 you'll be redirected to the Google sign-in page. After signing in you'll see the full app with the sidebar.

---

Usage

1. Sign in with Google
2. Click Upload PDF in the sidebar
3. Drag and drop a medical paper (PDF, max 50 MB)
4. Wait for indexing to complete (progress bar shows status)
5. Go to Chat and ask questions about the paper
6. Each answer shows:
   • The response with markdown formatting
   • A confidence score (high / medium / low)
   • Collapsible source snippets with relevance scores

---

API Endpoints

Method	Endpoint	Description
POST	/api/upload	Upload and index a PDF
POST	/api/query	Ask a question about indexed papers
GET	/	Health check

Example query

curl -X POST http://localhost:8000/api/query \
  -H "Content-Type: application/json" \
  -d '{"question": "What are the side effects of metformin?"}'

{
  "answer": "According to the uploaded paper, metformin commonly causes...",
  "sources": [
    {
      "text_snippet": "Gastrointestinal side effects including nausea...",
      "source_file": "clinical_trial_2024.pdf",
      "relevance_score": 0.91
    }
  ],
  "guardrails": {
    "confidence_score": 87,
    "is_supported": true,
    "warning": null
  }
}

---

Environment Variables

Backend (backend/.env)

Variable	Description
GEMINI_API_KEY	Google Gemini API key

Frontend (frontend/.env.local)

Variable	Description
AUTH_SECRET	Random secret for NextAuth session encryption
AUTH_GOOGLE_ID	Google OAuth Client ID
AUTH_GOOGLE_SECRET	Google OAuth Client Secret
AUTH_URL	App URL (http://localhost:3000 for dev)

---

Disclaimer

This tool is intended for research purposes only. It is not a substitute for professional medical advice, diagnosis, or treatment. Always consult a licensed physician.

License: MIT