Table 3 Comparison study on model inputs

From: Prediction of total and regional body composition from 3D body shape

Model Name

A

B

C

D

E

Model Inputs

–

–

–

SMPL

SMPL

 

Height, Weight

Height, Weight

Height, Weight

Height, Weight

Height, Weight

 

–

Waist

Waist, Hip

–

–

Method

Network

Network

Network

Linear

Network

Metrics

Fenland phase 2 R2

Total fat mass

0.884

0.909

0.910

0.897

0.922

Percentage body fat

0.739

0.792

0.797

0.766

0.823

Android fat mass

0.809

0.887

0.884

0.880

0.894

Gynoid fat mass

0.811

0.830

0.863

0.843

0.886

Visceral fat mass

0.698

0.779

0.792

0.774

0.802

Abdominal SCAT massa

0.701

0.727

0.730

0.716

0.723

Peripheral fat massb

0.802

0.824

0.832

0.821

0.872

Total lean mass

0.910

0.925

0.921

0.916

0.934

Appendicular lean massc

0.895

0.921

0.911

0.906

0.927

ALMId

0.739

0.853

0.824

0.833

0.853

  1. Best R-squared values are in bold font.
  2. aSCAT = subcutaneous adipose tissue.
  3. bPeripheral fat mass = arms + legs fat mass.
  4. cAppendicular lean mass = arms + legs lean mass.
  5. dALMI: appendicular lean mass index = appendicular lean mass/height2.
Back to article page