Public money buys this data. The public should get to read it.
To crossfoot a ledger is to add it across and add it down and check that the two totals agree — the oldest, plainest test of whether numbers hold together. That is what this project does to the healthcare system's own published figures, and where it gets its name.
Project Crossfoot collects the data that publicly funded healthcare is already required to publish, checks it the way an auditor would, and puts it where anyone can use it. The project is open source, and it is built to grow one analysis at a time.
Democratizing publicly funded healthcare data
Vast parts of US healthcare run on public money, and the law obliges the plans and programs that spend it to report how it is used — what gets approved and denied, how long decisions take, what happens on appeal, what care actually costs. In practice that reporting lands scattered across hundreds of payer and hospital websites, in formats no one opens, with nobody checking the arithmetic. Data that is technically public and practically unreadable is not public in any way that matters. Project Crossfoot exists to close that gap — and to close it in a way anyone can verify.
What is at stake
None of these numbers is abstract. In CY2025 the health plans in this dataset decided 73,167,403 requests for care and said no 7,752,713 times — each one a person whose treatment was delayed, changed or abandoned while paperwork moved. When someone pushed back, the insurer's own review reversed the denial 130,906 times; every one of those is a no that would have stood if nobody had checked. On the price side, the hospitals in the data bill $5.03 for every dollar that is actually paid, the same procedure carries wildly different prices depending on the building, and the county models on this site show what travels with the consequences — medical debt, and years of life lost before 75. Whether any of this is reasonable is a question for the public to argue. Project Crossfoot's position is only that the argument should happen over checked numbers.
Public in name only
The disclosure laws worked, in the narrowest sense: the documents exist. Then the trail goes cold. Prior-authorization metrics sit on hundreds of separate websites, some as scans, some as spreadsheets, some behind pages that only render in a browser. 125 of the 658 filings collected here publish no counts at all — whatever rate they claim, nobody can check it. 28 publish numbers that contradict their own arithmetic, and 342 hospital price files carry a price that contradicts another price in the same file. Nobody caught any of this, because compliance is measured by whether a file exists, never by whether it adds up. Transparency without verification is theater.
What one checked dataset changes
A patient deciding whether to appeal deserves to know that appeals succeed — the base rate is on this site, computed from the insurers' own filings. A journalist chasing a story needs more than an anecdote; here every claim traces to the payer's own document. A researcher or policymaker needs denominators, not press releases. An employer choosing a plan can look up how often it says no. And an insurer or hospital that believes its numbers are misread can point to the source document and the open code and show exactly where — that cuts both ways, and it is supposed to. You cannot argue with an anecdote. You can argue with a ledger.
The rules the work lives by
The same method holds whatever the subject, and every rule exists so that trust is never required — only checking.
Why now
Two things changed at once. The disclosure rules are new — hospital price transparency and the prior-authorization reporting mandate are both in their first years, which means the habits of this data are being set right now: whether anyone reads it, whether anyone checks it, whether publishing something that fails its own arithmetic carries any cost at all. And the tooling caught up — a crawler, a set of AI agents that read documents the way a patient clerk would, and validation code that never sleeps can now do what once took an institution. Project Crossfoot is a solo build, and that is part of the point: holding a regulated industry's numbers up to the light no longer requires anyone's permission. If the first years of this disclosure regime get audited — even by one person, in public, with the work shown — the next years get published more carefully. That is the theory of change, and it costs the public nothing to test.
An open project that wants company
Project Crossfoot is open source, and the work is sized for many hands — finding documents, checking numbers, and deciding what to count next. No healthcare background needed; the whole method is that the documents speak for themselves.
Built by Ryan Gomez
A technology leader, data architect and AI consultant in New York City, with 26 years spent turning intractable data problems into working systems — earning a patent along the way, and leading innovation for Fortune 500 enterprises and small family-owned businesses alike.
He runs Ryan Gomez & Co., an independent technology and AI consultancy, and has shipped this stack across healthcare, media, insurance and finance. Outside of work he served as a Rescue Task Force medic until retiring in 2026 — the same instinct he brings to data systems: stay calm, read the system, act.
Project Crossfoot is a solo build, and that is part of the point: the tooling to hold a regulated industry's numbers up to the light no longer requires an institution.
Ryan Gomez
- Base
- New York City
- Experience
- 26 years
- Practice
- Ryan Gomez & Co.
- Focus
- data architecture · AI