Inspiration
We were inspired by armchair physicians and how sometimes it can be hard to get medical advice/prognosis online.
What it does
Our program provides a user with a possible diagnosis given the symptoms they describe.
How we built it
We built the front end UI on Streamlit and we did the backend on Python along with using LangChain and Redis.
Challenges we ran into
We wanted to compare our results to MIMIC III data but we couldn't access the data as it is real clinical data and there are ethical issues that arise with such data. Additionally, we ran into some issues using AnyScale
Accomplishments that we're proud of
We are proud that the coding process went really well and we got it to work fairly quickly. We are also proud that we got to learn so much in the process.
What we learned
We learned how to use Streamlit along with how to use LangChain and Redis.
What's next for LLMD
We hope to be able to expand our medical database beyond just WebMD to get a better, more accurate diagnosis. Further more we need to scale it in the future using kubernetes.
Built With
- langchain
- llm
- machine-learning
- redis
- streamlit
Log in or sign up for Devpost to join the conversation.