For researchers

Medicare-Synth helps you explore ideas with artificial, Medicare-like datasets before working with restricted Medicare files. It is a research preparation and methods-testing sandbox—not a source of Medicare population estimates.

Choose your goal

Explore available data

Learn how beneficiary, enrollment, condition, and claims-style records relate, and what one row represents versus a patient count.

→ Understanding the data · What you can do

Plan an analysis

Walk the supported 2021 Original Medicare FFS-like diabetes and any-inpatient example: what fields exist, which public statistics were used as anchors, and what you can practice in code before restricted access.

→ Diabetes–hospitalization example · Getting started

Compare assumptions

See how synthetic datasets can match the same selected published rates yet differ on free assumptions (for example background inpatient risk). Use that to stress-test analysis code—not to recover “the” Medicare answer.

→ Interpreting results · Limitations

Before you go further

Not Medicare findings
Do not cite generated associations, model performance, or calibrated rates as empirical results about Medicare beneficiaries. Structural checks and agreement with selected published statistics do not authorize population inference.

Need setup commands, APIs, or contribution steps? Developer documentation →