We were SO over wading through 30 pages of confusing health insurance jargon. You know the feeling - endless documents that leave you more lost than ever about coverage and deductibles. That's why we built this app to decode your health insurance instantly.
Here's what it does:
Say goodbye to confusion: This app summarizes your policy in a snap, highlighting key factors you actually need to know. Compare with confidence: Pick two plans from different companies and see them side-by-side. No more guesswork, just informed decisions.
How'd we do it? We used React for our front end to create a user-friendly interface facilitating User details, file uploading, analyzing, and comparing insurance files. For the backend, Flask was employed to create endpoints to the database and the frontend to facilitate user login, insurance analysis, comparison etc. Through the information added about the user in the database we predicted the risk associated with that person. Used XG boost model to generate a model for the risk associated. We passed PDF's and stored the data in the form of chunks in a vector database, Pinecone, and leveraged RAG(Retrieval-Augmented Generation) with embedded queries through LangChain, to ensure tailored results for users. We used GPT3.5 Turbo augmented search functionality to deliver strict results based on queries and mitigated AI hallucinations and outdated information.
What we're most proud of? Using a vector database to implement a RAG and compare files along with using a sqlLite database to store user information and estimate the risk. Training the XG boost model to get the risk. This app goes beyond basic summaries. It delivers a deep analysis of your policies, empowering you to understand exactly what's covered.
But wait, there's more!
Next up, we're building a platform to process claims and get your funds faster. Imagine getting reimbursed without the hassle!
Built With
- flask
- javascript
- langchain
- openai
- pinecone
- python
- react
Log in or sign up for Devpost to join the conversation.