Rapid muscle volume prediction using anthropometric measurements and population-derived statistical models

Area of Science:

• Biomechanics
• Anthropometry
• Statistical modeling

Background:

• Muscle volume is a key indicator of muscle strength and power.
• Accurate estimation of fat-free muscle mass is crucial for various physiological assessments.
• Statistical learning offers novel approaches to predict body composition parameters.

Purpose of the Study:

• To develop and evaluate population-based statistical learning models for predicting lower-limb fat-free muscle volume.
• To identify key anthropometric predictors of muscle volume.
• To compare the efficacy of different statistical learning methods for this prediction task.

Main Methods:

• Computed tomography (CT) imaging data from 50 individuals were used to measure lower-limb muscle volumes.
• Six statistical learning methods were evaluated: stepwise regression, linear and polynomial SVM, and linear, quadratic, and spline-fit PLSR.
• Models were trained and tested using anthropometric measurements including sex, leg length, and shank girth.

Main Results:

• Linear PLSR achieved the highest prediction accuracy (87±2%) for fat-free muscle volume.
• Shank girth, sex, and leg length were identified as the most influential predictors.
• Support Vector Machine (SVM) performance varied significantly with interpolation method, while spline fit PLSR suffered from overfitting.

Conclusions:

• Linear PLSR models utilizing sex, leg length, and shank girth provide a reliable and practical method for predicting fat-free muscle volume.
• Stratifying populations by sex and adipose levels may enhance prediction accuracy.
• These findings support the use of non-invasive anthropometric data for estimating muscle mass.