Healthcare Price Transparency

• 

  README instruction when visit http://chat.3du.me:8000/ for the first time

What it does

Price transparency is a crucial issue in healthcare. It is essential for providers, payers, and patients to have access to healthcare prices and to be able to understand them, allowing individuals to make informed decisions about their healthcare choices. The Centers for Medicare and Medicaid Services (CMS) have defined three standardized machine-readable files for pharmacies, hospitals, and insurance companies.

Our Proof of Concept (POC) focuses on two key aspects: converting Hospital Price Transparency Machine-Readable Files into SQLite and developing a powerful retrieval-augmented generation (RAG) Chatbot. Our team manages and operates Price Transparency Products like TrueView and Pricing Data Lake, which provide a Web UI for searching shoppable services in hospitals. However, we identified a gap in the system - it doesn't support multiple languages, such as Spanish. It led us to envision a future where natural language interaction facilitates patient care cost estimation, transcending language barriers.

How I built it

I built it with LangChain, Chainlit and OpenAI API.

Firstly, we convert the machine-readable files into SQLite. Next, we use LangChain SQL Agent to query SQLite and integrate OpenAI GPT-3.5 for our Chatbot to understand and generate human-like responses.

Challenges I ran into

• I tried LocalAI with GPT4ALL-J model. But it can't provide the right response like OpenAI GPT-3.5+
• I need to find a way to translate the response based on input (question) language.

What I learned

• see https://jazzwang.github.io/cv/talks/2024-02-20_CGU_HPT_Chatbot.html#/14

What's next for Healthcare Price Transparency

Our future work involves leveraging LangChain to interface with OpenAPI functions, in line with our 'contract-first' strategy when developing Enhanced Eligibility solutions.

By combining technological prowess with a patient-centric focus, we aim to pave the way for a more accessible, understandable, and inclusive healthcare system.

Built With

• amazon-ec2
• chainlit
• gpt-3.5
• langchain
• openai
• python
• sqlite