Estimating Body Segment Properties for Adults AcrossDiverse Body Morphologies: A Data-Driven GeometricFramework

Technical Summary: Estimating Body Segment Properties for Adults Across Diverse Body Morphologies

Problem Statement
Accurate estimation of body segment inertial properties (BSIPs)—including segment lengths, masses, centers of mass, and moments of inertia—is a fundamental requirement for biomechanical analyses, musculoskeletal simulations, and clinical assessments. However, existing scaling methods are constrained by their reliance on limited datasets, which hinders their ability to generalize effectively across the full spectrum of diverse adult body morphologies. This limitation is particularly acute for individuals with higher body mass indices (BMI), where standard regression equations often fail to capture complex variations in mass distribution and geometry.

Methodology
The authors propose a data-driven geometric framework designed to estimate BSIPs using a combination of large-scale anthropometric datasets and regression modeling. The core methodology involves:

• Data Foundation: The models are trained on extensive datasets from ANSUR II and NHANES, providing a robust statistical basis for adult body variations.
• Geometric Representation: The human body is modeled as 16 distinct segments. The framework utilizes regression models to predict segment lengths, cross-sectional dimensions, and lumped masses.
• Input Variables: The primary inputs for the model are height, weight, and sex. To refine predictions for specific body shapes, the framework incorporates additional measurements such as waist and hip circumferences, or other length and cross-sectional metrics when available.
• Obesity Adaptation: A specific geometric rule set is implemented for individuals with obesity. This mechanism redistributes excess mass based on segment-specific volume changes, addressing the non-linear mass distribution often seen in higher BMI populations.
• Implementation: To ensure accessibility, the authors developed an open-source Python API that executes the full parameter estimation process using the trained models.

Key Results
The evaluation of the framework demonstrates that the resulting models successfully reproduce segment lengths, cross-sectional dimensions, and lumped segment masses within the ranges observed in the training datasets. Crucially, the framework outperforms previously published regression equations, with the most significant improvements observed in predictions for individuals with higher BMI values. The inclusion of circumferential measurements and the specific geometric rules for obesity significantly enhance the accuracy of shape and mass predictions compared to methods relying solely on height and weight.

Significance and Claims
The paper positions this framework as an accurate and accessible tool for estimating adult body segment properties across a wide range of body sizes and shapes. By leveraging large datasets and a geometric approach, the work aims to support improved motion analysis, musculoskeletal simulation, and clinical biomechanics. The authors emphasize the utility of the open-source API in promoting broad adoption, thereby addressing the gap in generalizability found in current scaling methods. The contribution is framed as a practical solution to enhance the fidelity of biomechanical models without requiring complex imaging for every subject, provided standard anthropometric data is available.