criticalaccess.aiAI for rural hospitals

Projects · Code on GitHub

Open tools and research, built on public data.

Each project starts from a question a hospital leader or a public health team asked. The code is on GitHub. Each page here says what the project does, what data it needs, and where it stops.

01 · Analysis and research

Questions answered with public data.

Two finished pieces of analysis and one study still in progress. None of them uses patient data.

  • ~/rural-hospital-outmigration-scan
    Analysis · public data only

    Where patients go, and what keeping them is worth

    Starts from where a county’s Medicare patients are admitted, then prices each way to keep care local under critical access payment rules. Produces a one-page briefing and a workbook that sources every figure.

    PythonNodeClaude CodeCMS public files
    Read about the project →
  • ~/hospital-website-audit
    Security and compliance audit

    Scanning hospital websites for hidden hacks

    A passive scanner that reads hospital websites the way a visitor does, looking for injected gambling and spam links and for hijacked former domains. Run across all 1,382 critical access hospitals on the roster.

    PythonInternet ArchiveDNS recordsMIT license
    Read about the project →
  • ~/hospital-hurricane-evacuation · in progress
    Research · not yet peer reviewed

    When sheltering in place fails

    A national study of hospitals in Atlantic and Gulf hurricanes since 1990, and of what happens when a hospital that stayed has to evacuate anyway.

    PythonpandasNOAA and CMS data
    Read about the project →

02 · Tools and prototypes

Things you can open and try.

A teaching simulation with a live demo, an AI agent with the tests that check it, and an education prototype. Each page says plainly what the tool is not for.

03 · How to read a project page

Every page says where the project stops.

A tool is only useful if you know what it can’t tell you. Each project page is laid out the same way, and the limits come from the project’s own documentation.

On every project pageFive parts, in this order

  • The question. What a hospital leader or a public health team wanted to know.
  • The data. Which public files it reads, and whether it needs anything of yours.
  • What it does. In plain steps, without the code.
  • What it found. With sample figures labeled as samples and unfinished research marked as unfinished.
  • Limits. What the project does not cover.

All of it is on GitHub.

Read the code, run it on your own county or your own website, and tell us what breaks.