Claim Decoder

Translating your medical bills from gibberish to English

Comment

Demo: https://www.loom.com/share/840c53b43bff4c448af458a060ef8914?sid=93ebe8c8-121a-4156-b78d-8fbb19c189e8

Inspiration

Ever received a medical bill and wondered “what the heck”? Medical billing is a $6 billion industry that baffles everyone. Patients can't understand their bills, and even professional coders and billers struggle with complex claims data. Take a look at this actual claims data:

ISA*00* 00 ZZ*ABCPAYER *ZZ*ABCPAYER *190827*0212^00501*191511902*0*P:~ GS*HP*ABCD*ABCD*20190827*12345678*12345678*X*005010X221A1~ ST*835*35681~ BPR*I*132*C*CHK*********456*20190331~ TRN*1*12345*1512345678~ DTM*405*20190314~ REF*EV*CLEARINGHOUSE~ N1*PR*DELTA DENTAL OF ABC~ N3*225 MAIN STREET~ N4*CENTERVILLE*PA*17111~ REF*2U*0101~ REF*HI*0202~ PER*BL*JANE DOE*TE*9005555555*EX*123~ PER*ICUR*myplan.com/policies~

What it does

Claim Decoder automatically translates medical claims from PDFs and EDI files into plain English that anyone can understand. Our platform flags questionable codes and billing amounts, then provides clear next-step recommendations for corrections. Whether you're a professional biller processing hundreds of claims or a patient trying to understand a single bill, we turn confusion into clarity. We're eliminating the confusion that costs healthcare billions in errors and delays.

How we built it

Built a JSON schema that captured all the useful EDI 835 data. Tested all the LLMs in parsing 835 EDI and pdfs to extract and transform data to put into JSON files (gemini 2.0 ended up being the sweet spot for integration complexity, performance, and accuracy). Integrated latest CMS 835 spec data and whatever we could find about payers to make this more reliable. Used v0 + chatGPT to build out a biller and provider facing front end. Web app is hosted on Vercel.

Challenges we ran into

Hello hallucinations!

Accomplishments that we're proud of

We successfully created a tool that bridges the gap between incomprehensible claims data and human understanding. Our platform empowers both professionals and patients to advocate for accurate billing, potentially saving thousands of dollars in errors and disputes.

What we learned

The medical billing industry desperately needs better tools, but the complexity runs deeper than we initially realized. We learned that effective solutions require not just technical translation, but also understanding the human frustration and financial impact of billing confusion. Also, there are 10,000 different billing PDFs between CMS and other official resources. What the heck!

What's next for Claim Decoder

  1. We want to improve recommendations for saving patients' costs. This requires including data from the No Surprises Act and Transparency in Coverage data, along with steps patients can take to bring down the cost of their bills.

  2. We plan to expand our payer database coverage, integrate with existing healthcare management systems, and develop predictive analytics to catch errors before claims are submitted. Our goal is to become the standard tool for claims clarity across the entire healthcare ecosystem.

  3. We plan to integrate with EHRs and clearinghouses to make this tool more useful for the billing community.

Built With

Share this project:

Updates