Medscribe

Inspiration

We were interested in finding solutions that would be helpful in other fields. A few members have family members that work in the medical field and we wanted to see if we could create something that would be useful for them.

What it does

The app is supposed to transcribe a conversation between a doctor/nurse practitioner and a patient. Based off the conversation, it would help generate soap analysis notes for the doctor when charting, and would also provide medication suggestions, possible medical billing codes, and document patient plans if the problem does not require medication.

How we built it

Frontend: React 19 + TypeScript + Vite
Backend: FastAPI + Celery (Python), AI pipeline via Gemini + scispaCy/medspaCy + ChromaDB (RAG)
Database: Supabase (Postgres + Realtime for live updates)

Challenges we ran into

We ran into a lot of problems finding appropriate api keys, connecting all of the components together. We ran into a lot of bugs, and because we don't have a lot of experience, we didn't really know how we needed to split roles, which made it difficult to know how we needed to work in it together.

Accomplishments that we're proud of

We were definitely proud with how far we got. With a string of failures and mistakes, we weren't sure how far we'd get.

What we learned

We definitely learned that we need to be more structured in the way that we approach this assignment. Although we had 2 weeks on this, we had to juggle working on homework, studying for test, and doing whatever else we needed to. If we were more efficient in the way that we approached this, we believe it would have turned out much better.

What's next for Medscribe

We're definitely going to keep working on this as we believe it's an interesting idea, though it'll definitely take sometime before it can actually work to the extent that he want it to work.