Where the data comes from

Public data, and anyone can reproduce it

There is no private or proprietary data behind the Atlas. Every person on the map comes from a free, public U.S. government health survey. The whole pipeline — download, combine, score, check — is scripted, so anyone can run it and get the same map.

The source: NHANES

The map is built on NHANES, the CDC’s National Health and Nutrition Examination Survey — one of the most trusted pictures of American health. It is already de-identified and released to the public. We use the 2013–2014 cycle, the most recent one that pairs lab work, a week of wearable-measured activity, and symptom questionnaires on the same people.

Our working sample is the 3,919 adults who have complete data across all three kinds of measurement — body, behaviour, and how they feel.

Exactly which pieces we use

Everything is joined per person on the NHANES respondent ID (SEQN):

Kind	What it is
Who	Demographics — age, sex, race/ethnicity, and the survey weights (file `DEMO_H`).
Body	Standard blood, metabolic, and inflammation labs, plus body measurements (`BMX_H`).
Behaviour	About a week of physical-activity-monitor data from a wrist device.
Feelings	Validated questionnaires on mood and self-rated health.
Conditions	Self-reported doctor-diagnosed medical conditions (`MCQ_H`) — used only to check the map, never to build it.

The raw files are downloaded straight from the CDC: the public NHANES 2013–2014 data page.

The pipeline, start to finish

Download the public NHANES files directly from the CDC.
Combine them per person on the respondent ID, keeping only adults with all three kinds of measurement.
Score each person’s burden from the combined signals.
Sort everyone into five equal burden groups.
Check the result against diagnosed conditions, with survey weighting and adjustment for age, sex, race and weight (see how we know it’s real).

Your privacy. NHANES is already anonymous public data. The “Where do I fit?” tool runs entirely in your own browser: it compares a few numbers against public statistics, once, and stores nothing. No account, no upload, no tracking of what you enter.

Reproducible by design

The download-to-map pipeline is committed as a small, self-contained project with its scripts, a data manifest, and the validation output. Because the inputs are public and the steps are code, the map is not a claim you have to take on faith — it is something anyone with the scripts can rebuild from scratch.

← The full map How we know it’s real → Where it’s going →