Inspiration

Our inspiration came from the growing problem of misinformation and public distrust in the scientific community. During the pandemic and beyond, we’ve seen how anecdotal evidence or weak studies can spread faster than validated research. We wanted to create a tool that helps people move from anecdotes to stronger, systematic evidence in a way that’s visual, understandable, and trustworthy.

What it does

The app will take in input generate a claim from either documents or plain text. We then generate a pyramid based off the evidence-based pyramid that doctors and professionals use with articles that reference the claim mentioned. We also provide a general summary of all the articles and provide an MLA format citation if someone wants to use the article in a paper.

How we built it

  1. We mapped out our project using Figma and Excalidraw to visualize user flows and interfaces.
  2. Development phase:
    • Started with Base44 to scaffold the application.
    • Ran into issues with API access, tokenization, and integration.
    • Tried post-processing Base44’s results, but eventually pivoted to generating a React app for more control.
  3. Iteration: We combined AI outputs with manual refinement, balancing automation with human oversight.

Challenges we ran into

Halluncinations, and code base access. API authorization and token count

Accomplishments that we're proud of

For standing up the application and solving the API json result using a white list.

What we learned

  • How to use the Base44 app for basic web development.
  • Where Base44 succeeds and where it struggles, requiring human intervention.
  • How to use Cursor in the development workflow.
  • The concept of “vibe coding” building quickly by intuition rather than overplanning.
  • The pitfalls of vibe coding, including messy integration and lack of structure. ## What's next for HealthCheck AI Want to generate a map of where claim misconception comes from. Fix some of the url problems we have.

Built With

Share this project:

Updates