What it does

AI agent to assist healthcare claim adjudicators, querying historical data to find similar claims to identify normal range of approved payment amount.

How we built it

Langchain with Llama and Claude models are used to call a custom "tool" we created. The "tool" queries the healthcare claims dataset based on a user input procedure code and to identify the normal range of approved payment amount based on historical data.

Challenges we ran into

  • new environment setup and admin activities
  • understanding new dataset
  • debugging streamlit and langchain errors

Accomplishments that we're proud of

  • end to end proof of concept in a short amount of time

What we learned

  • building agentic AI with tool usage capability
  • deploying Streamlit app

Built With

Share this project:

Updates