Predicting Body Segment Properties: Using Height, Mass and Gender to Estimate the Mass, Center of Mass, and Moment of Inertia of Body Segments

Abstract

Most of the current models used to investigate the biomechanics of human body do not account for variation in the population. Creating a more inclusive model is necessary to have better biomechanical analysis tools. By using anthropometric data to predict measurements using regression models, it is possible to estimate body segments masses, body segments volumes, centers of mass, and moments of inertia. These properties can be used to make models of various bodies. Regression models were created to estimate anthropometric measurements from height, mass, and gender. Two predictive models were developed, in parallel both with the same purpose. One model is based on the Bayesian modeling library Bambi and the other comes from the XGBoost library. Principal Component Analysis was used to investigate gender differences and to identify the most influential measurements. For the mass and inertial estimation part, the Modified Hanavan Model were used to estimate body segment mass, volume and moment of inertia. The body segment volumes were calculated using the 16 geometric shapes the Modified Hanavan geometric model consists of. By assuming uniform density the volume could be estimated and was validated by using the total body mass. Then the center of mass and the moment of inertia could be calculated using the volumes, masses and with formulas of the geometrical shapes. Together, anthropometric prediction and mass and inertial properties, gives the possibility to represent human variability in digital human modeling.