MediBot – The AI-Powered Medical Assistant

Inspiration

As healthcare becomes more digitized, doctors face a growing problem: information overload. Patient records are scattered, raw, and often hard to summarize quickly. MediBot was born out of the need to give medical professionals fast, clear, AI-powered insight into patient data.

What It Does

MediBot reads patient data, generates an intelligent summary using the Meditron AI model (via Ollama), then creates a narrated video using Tavus and ElevenLabs. Doctors can interact with the AI via chat for clarification, saving them time and improving decisions.

How We Built It

• Frontend: React + Vite for a fast and responsive UI.
• Backend: Node.js + Express to handle patient records and AI interactions.
• AI Engine: Meditron running locally via Ollama to generate clinical summaries.
• Media Generation: Tavus + ElevenLabs for voice/video.
• Storage: nft.storage + IPFS for decentralized patient record storage.
• Hosting: Netlify for frontend deployment.

Challenges We Faced

• Running Meditron locally via Ollama and handling long-form context inputs.
• Connecting AI outputs to Tavus/ElevenLabs for human-sounding narrations.
• Designing a UI that’s usable by doctors with limited time.
• Ensuring offline or local deployment was possible for data-sensitive regions.

What We Learned

We learned to integrate multiple AI services seamlessly, use local LLMs for sensitive data, and provide real-time medical insights that could scale globally. MediBot is our first step toward smarter, AI-first healthcare.

Built With

• api
• dataset
• elevenlabs
• express.js
• json
• local
• meditron(ollama)
• node.js
• react
• tavus
• vite