Researcher FAQ
Are these real Medicare patients?
No. Records are artificial and Medicare-like. The project does not reconstruct real beneficiaries.
Do I need CMS permission or files to try the demos?
Not for the built-in structural fixtures. Those run from the open-source package without bundling restricted Medicare claims. If you later acquire official CMS synthetic source files for other workflows, check that product’s acquisition and redistribution terms separately. Code is Apache-2.0.
Can I use it without coding?
You can read the researcher pages, worked example, and interpretation guides without coding. Producing tables still needs a developer path or a collaborator who exports files. There is no browser-based generator or one-click calibrated download on the website today.
Does it support my disease or year?
The calibrated MVP is pinned to a 2021 Original Medicare FFS-like, age 65+, diabetes and any-inpatient context (with heart failure as an indicator). Arbitrary disease/year/outcome calibration is not a current product.
Can I publish analyses of these data?
You can publish methods, software, and clearly labeled synthetic demonstrations. Do not present synthetic associations or calibrated rates as empirical Medicare findings. Cite limitations and evidence anchors.
How realistic are admissions and diagnoses?
Structural relationships are checked. Selected summary rates can be targeted in the MVP. Realistic longitudinal patterns, diagnosis-specific admissions, rich comorbidity dependence, and validated subgroup behavior are not claimed. Phase 6 evaluation remains active.
Can I use synthetic results to justify a clinical conclusion?
No. That is not a valid use.
Where do I find setup and technical details?
Start at the developer hub, especially
installation,
CLI, and
calibrated worlds.
Canonical status lives in
ROADMAP.md
and SPEC.md.