# Table 2 Prediction equations for ALM developed on 2/3 of the sample and internal cross-validated on the remaining 1/3a

Validation for BIA-based ALM prediction equation
Development group (n = 131)
Measured ALM = 17.28 ± 3.62
Predicted ALM = 0.254ZI@2 MHz + 1.0324sexM1F0 + 0.066Xc@5 kHz + 1.396
R = 0.964, Adusted R2 = 0.928, SEE = 0.97 kg, CV = 5.6%, SR: Excellent (M), Very good (F),
VIF: ZI@2 MHz = 3.39, SEXM1F0 = 3.46, Xc@5 kHz = 1.11
Predicted ASM = 17.25 ± 3.48
Internal Cross-validation group (n = 64)
Measured ALM = 17.77 ± 3.90
Predicted ALM = 17.68 ± 3.75, p = 0.486*
R = 0.965, R2 = 0.931, TE = 1.00 kg, CV = 5.6%, SR: Excellent (M), Very Good (F), PIA: 84.4%
1. ZI@2 MHz = impedance index at 2 MHz; SEXM1F0: man = 1, women = 0; XC@5 kHz = reactance at 5 kHz; SEE (Standard Error of Estimate) =$$\sqrt{\sum {(Measured ALM-Estimated ALM)}^{2}/(N-p-1)}$$where p = number of predicter variables); SR = subject rating [ideal = 0.72 ~ 0.90(M), 0.54 ~ 0.65(F); excellent = 0.90 ~ 1.09(M), 0.65 ~ 0.83(F); very good = 1.09 ~ 1.27(M), 0.83 ~ 1.01(F); good = 1.27 ~ 1.45(M), 1.01 ~ 1.16(F); fairly good; fair; poor] ; TE = $$\sqrt{\sum {(Measured ALM-Estimated ALM)}^{2}/N}$$; PIA = the number of individuals within ± 1.45 kg ALM for men and ± 1.16 kg ALM for women of difference between predicted and measured ALM × 100 / total), * = paired t-test betwee measured ALM and predicted ALM 