Unflame
Back to methodology

Onboarding calculation

Body Fat and Lean Mass Methodology

How onboarding resolves body fat percentage, lean mass, fat mass, and body surface area.

Status: Used by calorie, protein, hydration, fiber, and activity target calculations.
Last updated: June 15, 2026

Inputs and outputs

Inputs

  • Age
  • Gender selected in onboarding
  • Weight in kilograms
  • Height in centimeters
  • Optional body-fat estimate or skinfold measurements

Outputs

  • Body fat percentage
  • Lean mass in kilograms
  • Fat mass in kilograms
  • Body surface area

Formula

fallback_body_fat = age_gender_average_table(age, gender)
visual_body_fat = male ? 5 + slider_pct * (40 - 5) : 12 + slider_pct * (50 - 12)
skinfold_sum = biceps + triceps + subscapular + suprailiac
L = log10(skinfold_sum)
body_density = Durnin_Womersley_coefficient(age, gender, L)
skinfold_body_fat = (495 / body_density) - 450
resolved_body_fat = clamp(chosen_or_fallback_body_fat, 3, 60)
lean_mass_kg = weight_kg * (1 - resolved_body_fat / 100)
fat_mass_kg = weight_kg - lean_mass_kg
body_surface_area = 0.007184 * height_cm^0.725 * weight_kg^0.425

Calculation steps

  1. Use the user-provided or scan-estimated body fat percentage when present.
  2. If no body-fat estimate is available, use the app demographic fallback table by age and gender.
  3. For manual caliper mode, use the four-site Durnin-Womersley skinfold density table and convert density to body-fat percentage with the Siri equation.
  4. Clamp impossible values, then compute lean mass and fat mass from body weight.
  5. Compute body surface area with the Du Bois formula for context.

Guardrails

  • Body-fat estimates are planning estimates, not a diagnostic body-composition test.
  • The demographic fallback is intentionally treated as lower confidence than a direct measurement.
  • The visual slider is a rough estimate and should not be treated as a medical measurement.

Sources

These formulas describe how Unflame estimates onboarding targets. They are planning estimates and should be reviewed against real-world trends, user preference, symptoms, and professional guidance where relevant.