How it works

From a recording to a number you can check.

How data gets in, what each value is called, and what stays yours.

1 · Data sources & importing

Three ways in.

Automatic device sync

Not available in the public app.

Automatic sync from devices, including Garmin, is not available to public users. You use manual entry and supported file imports.

File import

You bring the files.

VITAL reads the file formats below. “File import” means exactly that: VITAL reads the file you give it. It does not connect to the service that produced it.

Manual readings

Type it in.

Enter a reading with an optional note. If you correct it later, the history of the correction is kept. These values are labelled Reported by you.

File formats VITAL can read

  • FITActivity files from watches and bike computers. Every session in a multi-session file becomes its own activity.
  • TCXActivity files with laps and heart rate.
  • GPXTracks with positions and, where present, heart rate.
  • Apple Health exportexport.zip or export.xml: resting heart rate, HRV (SDNN), VO2 max, sleep, workouts, weight and blood pressure.
  • CSVWeight or blood pressure. Units are never guessed: a file without them is rejected.

What happens on import

  • Likely duplicates are flagged, never deleted. A session that starts within 120 seconds and has a distance within 5% of one you already have is flagged. You choose: keep the existing one only, or keep both.
  • The same file twice is recognised by its checksum and skipped.
  • Rejected records are counted by reason and shown to you, rather than quietly dropped.
  • Imported series stay separate. Apple heart-rate variability (SDNN), for instance, is its own trace and is never merged with Garmin overnight HRV, which measures something different.

Provenance and storage

  • Every figure keeps the provider it came from, and each import records the file name, checksum, parser version and time.
  • The uploaded file is staged briefly while it is processed and is not kept afterwards. Only the parsed result and file details remain.
  • Existing data is never modified by an import.

How well is this tested? Most import formats have been tested against synthetic files, with a small set of files from real devices also verified. Apple Health imports have so far been tested only against synthetic exports. Check imported results against your device.

2 · Recorded, calculated, estimated

The label comes first.

Every value belongs to exactly one evidence class, and the page makes that class visible before it tells you what the number means.

  • Recorded

    A recorded fact. As measured by the device, or entered by you. A heart-rate reading, a step count.

  • Calculated

    A derived measurement. A deterministic calculation from recorded facts — a 7-day median, a pace from distance and time.

  • Estimate

    A supported estimate. An inference that states its evidence and its uncertainty — VCI, or a reconstructed pace line.

  • Reported by you

    Your own words and numbers. Manual readings, perceived effort, notes. Shown as yours, not as measured.

  • Interpretation

    A possible reading, worded as possible.

  • Can’t tell

    Said plainly, with what is missing.

3 · Baselines & longitudinal analysis

Your own earlier readings are the reference.

A baseline is your own earlier readings over a stated window, built from the median and its spread. Missing days are never filled in. Windows by page:

Command tiles
The 7-day median against your preceding 30 days, unless the tile says otherwise.
This week vs last
Observed totals and means from your recordings, this week against the week before.
Health: your usual range
Your latest 7-day average against your own variation over the 60 days before. For weight and blood pressure: the latest 14 days against the 90 before.
Recovery lanes
The last 14 days against the reference period you set for the journey. A recent-window comparison, not a verdict on the whole history.
Trends
You choose the range and the baseline window. Smoothing is a trailing mean drawn over the raw points; it never feeds into baselines or exports.
VCI
Sessions from the previous 90 days. The headline is the best level you have demonstrated; only repeated, credible evidence lowers it.

Gaps stay gaps.

A shaded stretch means no readings. The line is never drawn through it as zero, and a gap is not a change.

Enough data, or no verdict.

A comparison needs a minimum number of readings or days in each window. Below it, VITAL shows the counts and no verdict.

Higher or lower.

A change is shown as higher or lower, not labelled good or bad. Context decides what it means, and you hold the context.

4 · Evidence & uncertainty

Know when the evidence runs out.

The standard VITAL is built to: “Never manufacture certainty. Never optimise a result to make a graph attractive.”

Sample counts and windows, on the page.

Comparisons show how many readings they rest on and which window they cover, so you can judge how much weight to give them.

Missing data is announced.

Where data is missing VITAL says so rather than estimating. An empty page says what is missing and what would fill it.

Associations are not causes.

Exploratory pairings show their sample counts and limitations, and are never presented as one thing causing another.

A good-looking graph is not proof.

Two numbers from the same source do not confirm each other. VITAL looks for an independent check before treating agreement as evidence.

Not medical software. VITAL is personal health intelligence: it shows how your own recorded data compares with your own history. It does not diagnose, treat or monitor any condition, and it is not an emergency service.

5 · User control

Your recordings are yours.

Originals are preserved
VITAL does not alter your recorded measurements. Calculations and estimates are shown separately and labelled as such.
Export everything
One zip with a JSON file for every table of your own database, plus your account, consent and audit records, and a README.
Delete your account
Typing DELETE MY DATA removes your database at once. This cannot be undone. The privacy page lists what remains afterwards and why.
Consent is recorded
You give two consents before using VITAL: that you have read the privacy notice, and that VITAL may store and analyse the data you import for your own use. Withdrawing consent closes your data routes immediately.
Sharing is opt-in
Groups share nothing until you choose a category in a group, and you can stop at any time.

Ready when you are.

VITAL is free to use, with no ads. Already a member? Sign in. New to VITAL? See how to get an account.