Skip to main content

About the project

🏮 Lantern: A Light Carried Through the Hospital Corridor
Lantern is an AI-powered reading aid designed to help patients—especially those who speak English as a second language—understand dense, complex hospital appointment letters. By photographing or pasting a letter, the app extracts crucial instructions, fasting rules, timed steps, and medication details, transforming them into an accessible, colour-coded timeline, fasting ring, voice walkthrough, and medicine flash cards with visual body maps.

To prevent dangerous translation errors (such as wrong times, altered dosages, or swapped left/right instructions), Lantern checks every translated line using deterministic code and a second independent AI model. If a line fails verification or cannot be checked, Lantern falls back to showing the original English text alongside a warning and the hospital’s phone number—and it will never read that unverified line aloud. Built with Next.js and Google Gemma 4, Lantern prioritises safety, patient trust, and accessibility without making medical guesses or providing medical advice.

Built with

claudegemmaollamagemini

AI tools

Coded withClaude CodeGoogle AI StudioClaude Sonnet 5.5

Prize Categories Entered

  • Deepmind Google

    Best Use of Gemma 4

    Submitted
  • DigitalOcean

    Best Open-Source AI Project

    Submitted
  • Event prize

    Spooky Challenge

    Submitted
  • Event prize

    Main theme - Fusion

    Submitted