Personalized Tibial Strain Prediction: Feasibility of a Novel Approach Using Markerless Motion Capture and Statistical Shape Models

This study introduces a novel framework for generating personalized musculoskeletal models to predict tibial strains from video data. By integrating a statistical shape model (SSM) with markerless motion capture, this approach enables strain prediction without requiring medical scans or marker-based data collection, making it particularly useful in settings like basic combat training. Additionally, we evaluate the impact of using single versus multiple principal components in estimating musculoskeletal injury risk indicators, such as tibial strains. Data from seven participants performing a one-legged hop were used to evaluate this framework. To determine the impact of using single versus multiple principal components (PCs) on tibial strain predictions, the same loading profile was applied to the personalized models. Based on the observed effects, we selected the appropriate number of PCs and applied personalized loading profiles to the corresponding finite element (FE) models to predict tibial strains. Our findings indicate that using only the first PC to develop FE musculoskeletal models leads to differences in strain predictions compared to models incorporating multiple PCs, as the latter capture subtle morphological variations. The first PC primarily accounts for variability in overall size and relying solely on it may result in similar strain predictions for subjects of comparable stature. This study highlights the importance of incorporating higher-order PCs to better capture morphological differences, ultimately influencing strain distribution and injury risk assessment. The approach presented in this study offers a scalable, efficient solution for personalized biomechanical modeling, with applications in both individual and population-level injury risk analysis.