Development of a new formula for estimating human body mass using data from Size Stream SS20 3D body scans. These formulas were developed and cross-validated using a diverse set of over 1,790 human 3D body scans (of 179 individuals). Machine learning techniques were utilized to correlate the 3D body scan data to calibrated body mass measurements.
The new formulas yield impressive results with R2 (R SQUARED) > 0.95 in terms of kilograms body calibrated scale data and the estimate from the 3D body scan. The STD is 3.47.
Key takeaway
Size Stream’s body mass formula achieves R² > 0.95 using just four variables, gender, stomach circumference, right thigh circumference, and total body surface area, the last of which is automatically generated by every SS20 scan but impossible to replicate with a tape measure. Developed at Texas Tech University and cross-validated at Pennington Biomedical Research Center and UCSF/University of Hawaii, it carries the same multi-site rigor applied across Size Stream’s full formula suite. Combined with body fat and lean mass formulas, a single SS20 scan now delivers a comprehensive body composition profile, weight, fat, lean mass, and hundreds of measurements, from one frictionless imaging session, with or without a scale.