Identify demographic representation of widely used OpenSim lower-limb models

Determine which age and sex demographic groups are adequately represented by the widely used open-source OpenSim lower-limb musculoskeletal models—Delp et al. 1990 (Model D), Gait2392/2354 (Model G), Arnold et al. 2010 (Model A), London Lower Limb model (Model L), and Rajagopal et al. 2016 (Model R)—by quantitatively comparing each model’s musculotendon parameters and physiological cross-sectional area (PCSA) distributions across lower-limb muscles to experimentally measured values for females and males and for young adults, adults, and elderly cohorts.

Background

Scaled generic musculoskeletal models are commonly used in OpenSim, but maximum isometric forces and related PCSA parameters are often left unscaled and applied uniformly across populations. Model outcomes are highly sensitive to the relative distribution of maximum isometric forces across muscles, and demographic differences in muscle strength distribution can substantially affect predictions.

The paper compiles experimental PCSA and muscle mass data and shows that current open-source models inconsistently represent age- and sex-specific muscle distributions. The authors state that it remains unclear which demographic groups (by age and sex) these frequently downloaded models actually reflect, motivating a precise mapping from model parameters to empirically observed demographics.