Where does fat mass come from?

Sahha derives it as the day's latest weight multiplied by the day's latest body fat percentage. It needs both readings on the same day from the same source, which in practice means a smart scale.

Is lean mass derived the same way?

No. Lean mass is read directly from the platform's lean body mass record, which scale apps write. It is not computed as weight minus fat mass.

Why does body fat require a wearable when weight does not?

Because only a bioimpedance scale or a similar device can estimate it. A manual entry is possible but rare.

Body fat | Sahha

What the biomarkers measure

Biomarker	What it is	Unit
`body_fat`	The latest body fat percentage reading on the day	Percent
`fat_mass`	Latest weight × latest body fat percentage, on the same day	Kilograms
`lean_mass`	The latest lean body mass reading on the day	Kilograms

Field	`body_fat`, `fat_mass`, `lean_mass`
Unit	Percent, kg, kg
Reported	On days with a reading. The latest reading of the day is kept.
Window	Midnight to midnight, profile local time
Wearable required	Yes, a body-composition scale or similar
API	`GET /api/v1/profile/biomarker/{externalId}?categories=body` plus one `types` parameter per field and a date range
Webhook	`BiomarkerCreatedIntegrationEvent`. Entries update in place, upsert on `id`.

Like weight and height, these are point-in-time readings with periodicity none. Take the most recent entry as the current value.

Data sources

Platform	Body fat	Lean mass
iOS, HealthKit	`bodyFatPercentage`, from smart-scale apps such as Withings, Renpho, Garmin Index, and Eufy	`leanBodyMass`, from the same apps
Android, Health Connect	`BodyFatRecord`, from scale apps and Samsung Health	`LeanBodyMassRecord`
Direct integrations	Provider-dependent	Provider-dependent
Phone only	Manual entries only	Manual entries only

How Sahha calculates body fat

Pick one source for the day within the body category. A source that reports body fat ranks first, so the scale wins over any manual weight entry.
Body fat and lean mass are the latest reading of each on the day from that source.
Fat mass is derived: the day’s latest weight from the winning source multiplied by the day’s latest body fat percentage, divided by 100. It is written only when both exist on the same day.

Absence means no reading that day, or for fat mass, no day with both inputs.

Interpreting the value

Consumer body fat estimates are imprecise and consistent. Bioimpedance scales can be several percentage points off a reference method [1], but they drift little from day to day for the same person under the same conditions. The trend over weeks is meaningful; the absolute value is a rough guide.

Hydration moves the reading. Morning, post-workout, and evening readings differ. Coach users to weigh at the same time of day.

Lean mass is the number strength-focused users want. Fat mass falling while lean mass holds is the pattern a body-recomposition feature should look for.

None of the three is a score factor or has a trend or comparison insight.

Use cases

A composition trend. Plot fat mass and lean mass over 90 days as two lines, with weight as their sum. The identity holds whenever all three come from the same day.

GET /api/v1/profile/biomarker/{externalId}?categories=body&types=body_fat&types=fat_mass&types=lean_mass&startDateTime=2026-06-05&endDateTime=2026-09-03

A same-scale rule. Compare readings only within one source. A user switching scales will see a step change in body fat that has nothing to do with their body.

Limitations

Scale-dependent. No scale, no data. Coverage is a small fraction of profiles.
Fat mass needs same-day inputs. A weigh-in without a body fat reading produces no fat mass.
Point in time. Do not aggregate across days.
Absence is not zero.

Weight, height, and BMI for the readings body fat is combined with
Waist circumference for a composition proxy that needs no scale
Data dictionary in the developer docs

References

Achamrah N, Colange G, Delay J, et al. Comparison of body composition assessment by DXA and BIA according to the body mass index: a retrospective study on 3655 measures. PLoS ONE. 2018;13(7):e0200465. https://doi.org/10.1371/journal.pone.0200465

Body fat