AI Health Assistant RAG Application

Multiagent healthcare assistant routes patient queries to specialized AI agents (Symptom, Medication, Lifestyle) using Gemini Flash, with smart appointment booking and conversation memory.

Comment

🔧 Update: Python Version

project uses Python 3.9.6.

🏥 AI Health Assistant RAG Application

Short Description (25 chars left):
Multiagent healthcare assistant routes patient queries to specialized AI agents using Gemini Flash, with smart appointment booking and conversation memory.

📋 Problem Statement & Solution

  • Problem: Millions search online for health advice but lack immediate, safe, and context‑aware guidance.
  • Solution: A multiagent system that routes queries to symptom, medication, or lifestyle specialists, maintains conversation history, and books appointments and all privacy‑first (no accounts, contact only at booking).

🛠️ Technical Implementation

  • Backend: FastAPI (Python 3.9.6)
  • Orchestration: Router agent using Google Gemini Flash for intent classification
  • Specialist Agents: Three domain‑specific agents with engineered prompts
  • Session Manager: Stores conversations as JSON (sliding window of last 5–10 messages)
  • Appointment System: Keyword‑based specialist matching (specialist_mapping.json) and booking with contact capture
  • Frontend: Vanilla JS chat interface served from /chat
  • Testing: 50 pytest tests covering API, session management, and booking flows

🌐 Code Repository

🔗 Link: https://github.com/JaveriaBaloch/AI-Chatbot-Challenge

🎥 Demo Video / Presentation

(Optional but recommended)
A short video (2–3 min) showing:

  • Starting a chat, asking about a symptom
  • Router selecting the correct agent
  • Booking an appointment flow
  • Viewing chat history

You can use Loom or Google Slides with screen recordings.

📦 Tools, Datasets & Frameworks

  • APIs/Frameworks: Google Gemini Flash, FastAPI, Pydantic, Pytest, Uvicorn
  • Languages: Python 3.9.6, JavaScript (vanilla), HTML/CSS
  • Data: specialists.json, specialist_mapping.json, appointments.json (all locally generated/managed)
  • Libraries: python-dotenv, pytest, etc. (see requirements.txt)

✅ Guidelines Compliance

  • Originality: Built from scratch during this hackathon.
  • Team Size: Individual.
  • Third‑Party Libraries: Listed above; Gemini API is the key external service.

🚀 Quick Start

# Set up environment
echo "GEMINI_API_KEY=your_key" > .env
cd backend
python3 -m venv venv
source venv/bin/activate  # or venv\Scripts\activate on Windows
pip install -r requirements.txt

# Run tests
pytest test/ -v

# Start server
uvicorn main:app --reload --port 8000

Access the chat UI at http://localhost:8000/chat.

Share this project:

Updates