AI & Machine Learning / CASE STUDY
Aura Health.
AI-Powered Health Awareness Chatbot
01 — THE CHALLENGE
The problem.
Finding reliable health information online can be difficult. Health websites often contain long and technical information, while generic AI chatbots may give incorrect answers. Language can also be a barrier for people who are more comfortable using Hindi or Hinglish.
02 — THE APPROACH
A thoughtful solution.
Built Aura Health, a health awareness chatbot that uses trusted and admin-approved health sources to answer user questions. The system collects information from selected sources, cleans and stores it for search, then finds the most relevant content before generating an answer. It also supports English, Hindi, and Hinglish, detects urgent queries, shows the sources used for answers, and provides private counselor booking.
03 — BEHIND THE BUILD
How it came together.
The main challenge was making a health chatbot useful without allowing it to freely guess medical information. To solve this, the project uses a source-based approach. Admins first add trusted health websites to the knowledge base. The system then extracts the content, removes unnecessary text, breaks it into smaller sections, creates searchable representations, and stores them for later use. When a user asks a question, the system first finds relevant information from this knowledge base and then uses that information to prepare the response. This makes the answers easier to trace back to their source. The project also adds multilingual replies in English, Hindi, and Hinglish, an emergency query detector, Google sign-in, and confidential counselor booking. The result is a practical health-awareness platform that focuses on clear information and knowing when to ask for human help.
04 — THE RESULT
Source-Based Multilingual Health Assistant
Created a single platform where users can ask health-related questions in simple language, get answers based on trusted sources, check where the information came from, and request confidential counseling when needed.
Screens, systems,
and the details.
06 — DEVELOPMENT MILESTONES
The build, step by step.
- 01
Built the React frontend and FastAPI backend with Supabase authentication and database integration
- 02
Created the health knowledge pipeline for scraping, cleaning, chunking, embedding, and storing trusted source content
- 03
Added RAG-based question answering with Mistral AI, multilingual responses, and source transparency
- 04
Added emergency query detection, admin tools, analytics, and confidential counselor booking
- 05
Deployed the frontend on Vercel and backend on Render
07 — LESSONS & CONSTRAINTS
What made it interesting.
Making health answers depend on trusted source content instead of unsupported AI-generated information.
Cleaning and organizing health information collected from different websites.
Supporting English, Hindi, and Hinglish while keeping responses easy to understand.
Handling sensitive and urgent health questions safely.
Connecting authentication, database, AI processing, and the frontend into one working application.