Lean soft tissue and appendicular lean soft tissue in adolescent soccer players: new predictive equations using raw multifrequency impedance analysis data.

BACKGROUND

As determinants of soccer player performance, lean soft tissue (LST) and appendicular lean soft tissue (ALST) require accurate, segment-specific assessment.

OBJECTIVE

This study aimed to develop LST and ALST predictive equations from multi-frequency bioelectrical impedance analysis data at 50 and 500 kHz using raw data from an octapolar device and evaluate their predictive performance using dual-energy X-ray absorptiometry as a reference method via split-sample validation.

METHODS

Multiple linear regression models were developed using body mass, impedance index, and maturity offset as predictors. Model performance was evaluated using adjusted R², standard error of estimate (SEE), root mean square error, mean absolute error, Lin's concordance correlation coefficient (CCC), and Bland-Altman agreement analysis. Participants were randomly allocated into a development (n = 100) and an independent validation subset (n = 49).

RESULTS

Predictive performance was comparable between frequencies, with high explanatory power (R² for LST = 0.953-0.955; ALST = 0.931-0.932) and low prediction error (SEE ≤ 1.65 kg). Maturity offset contributed to LST models only at both frequencies (P < 0.001). In the validation subset, concordance was higher for LST (CCC = 0.973 at 50 kHz and 0.974 at 500 kHz) than for ALST (CCC = 0.955 at both frequencies), whose lower confidence bound fell within the moderate range. Absolute limits of agreement were narrower for ALST, but relative limits of agreement were larger for ALST (∼7.4-7.7%) than for LST (∼5.5-6.0%), and proportional bias was detected in ALST models.

CONCLUSIONS

These equations show strong potential for group-level estimation in adolescent soccer players. Individual-level application should consider this direction-dependent error, particularly for ALST, and longitudinal monitoring requires further validation.