MediCompanion

Inspiration

Healthcare systems around the world are currently overwhelmed, underfunded, or inaccessible to large populations who need health care the most. We believe that everyone deserves a right to personalized and accessible healthcare, inspiring MediCompanion.

What it does

Uses a local LLM to preserve context on a user's personal health profile and makes personalized suggestions based on this updated context. Ie, can recommend OTC drugs based on allergies or other health conditions, or providing an updated Medical ID.

How we built it

We used Django for the backend, React for the front end, as well as python scripts for webscraping and a Llama 3.2 1b local LLM model.

Challenges we ran into

Creating prompts that were specific enough to get consistent and reasonable answers from our local LLM was extremely difficult. Especially when taking into account the run time of it all. We came up with a not-so-perfect solution, but our main goal was just to make a proof of concept and we did that quite well.

Accomplishments that we're proud of

Finishing the project is our greatest accomplishment this weekend. Even if we don't win, creating something as a team is something we can all be proud of.

What we learned

Everyone was able to learn some new technology. Whether its React for our newer hackers or LLM prompt engineering for our seasoned veterans, everybody had something to learn.