Methodology & Published Sources
Effective date: July 17, 2026 · Last updated: July 17, 2026
This page explains, in plain language, how LeanSignal turns raw measurements into interpretations — and where its methods come from. Published equations inform LeanSignal’s estimates; they are not medical validation of LeanSignal as a whole.
What the Signal Score represents
The Signal Score expresses how strongly your own measurement history supports a repeatable pattern — for example, a genuine downward trend in weight — versus ordinary measurement noise. A high score means the evidence is consistent across time and, where available, across methods. A low score means the change is still within the range your measurements normally wander.
The Signal Score does not represent a clinical measurement, a body-composition ground truth, or proof that a specific tissue change occurred. It is confidence guidance about patterns in your data, nothing more.
Personal baseline and noise model
Everyone’s measurements bounce differently. LeanSignal watches your own history to learn how much your readings typically vary from day to day — your personal noise model — and uses that, rather than population averages alone, as the yardstick for what counts as a meaningful change. This is why the Signal Score “builds” over your first measurements: the baseline has to be earned before it can be trusted.
Why period averages
Single readings are unreliable — hydration, meals, time of day, and technique can each move a reading by more than the change you care about. Averaging over periods (weeks, months) suppresses this noise, which is why LeanSignal’s reviews and comparisons are built on period averages rather than single data points.
Combining body-composition methods
Where multiple applicable methods exist — for example, a smart-scale body-fat estimate, a tape-measure circumference method, and skinfolds — LeanSignal combines them, weighting each by how reliable that method tends to be and how well it agrees with your other data. When methods agree, the combined estimate narrows. When they disagree, LeanSignal widens the working range instead of picking a winner — disagreement is real information about uncertainty.
DEXA calibration
A DEXA scan can anchor your other estimates, and LeanSignal treats it as a calibration reference — but calibration is source-specific: a scan calibrates estimates relative to that DEXA source, not every method universally, and DEXA itself is not perfect ground truth.
Why forecasts remain conditional
Forecasts project your recent pattern forward, with a range that reflects your personal noise. They are conditional on your circumstances staying broadly similar — a change in routine, diet, training, or measurement habits can move you outside any forecast. Forecast ranges are practical uncertainty guides, not clinical confidence intervals or promises.
Published foundations
LeanSignal’s body-composition estimates draw on long-established, published anthropometric equations and measurement research, including the families of methods associated with:
- Jackson & Pollock skinfold equations (men and women).
- Durnin & Womersley four-site skinfold method.
- U.S. Navy circumference-based body-fat method (Hodgdon & Beckett).
- Siri and Brozek body-density conversion equations.
- Published research on day-to-day body-weight variability and measurement error in consumer body-composition devices.
The exact studies referenced by the app are listed inside LeanSignal alongside the features that use them. Published equations describe population relationships with known error; they inform LeanSignal’s estimates but do not validate the app as a medical tool.
What remains proprietary
The way LeanSignal blends methods, maintains the personal noise model, scores signal strength, and decides when to widen or narrow working ranges is NovoBuild’s own interpretation engine and is not published. The inputs (your measurements) and the foundations above are open; the blending recipe is proprietary.