Extending Musculoskeletal Modelling Validation Beyond Gait: Knee Loading and Kinematics During Daily Activities

Abstract

Musculoskeletal (MSK) modelling enables non-invasive estimation of internal body loads, which could inform the design of novel orthopedic implants and guide personalized treatment strategies. Extensive validation is a pre-requisite for the translation of MSK-modelling frameworks into clinical applications. This study validates predicted knee loads and local kinematics across multiple activities of daily living (ADLs) using motion-capture data from 184 activity cycles collected in six subjects performing five ADLs from the CAMS-Knee datasets. Personalized models incorporated subject-specific anthropometric, radiographic, and implantation data, along with a 6-DOF elastic-foundation contact model between implant components and a surgically balanced ligament model derived from cadaveric testing. Across activities, predicted knee contact force magnitudes showed good agreement with instrumented implant measurements (RMSE 0.46 BW; R2 = 0.81). The highest accuracy was found during level walking, whereas stair descent exhibited the largest discrepancies. Predicted implant kinematics showed qualitatively similar trends to fluoroscopic measurements across all ADLs, although with varying levels of accuracy across different degrees of freedom. EMG comparisons revealed broadly consistent activation timing between model predictions and experimental data, although gastrocnemius activation was poorly estimated during stair descent and downhill walking. By validating MSK model predictions across the complete range of activities available in the CAMS-Knee datasets, this study extends validation beyond previously analysed gait or bending tasks and reveals activity-specific discrepancies that reflect current limitations in modelling assumptions and would remain undetected in single-activity analyses. These findings provide a robust basis for future refinement of MSK models and support their development toward clinically meaningful applications.