perform biomechanical analysis

This work addresses the challenge of deploying monocular 3D biomechanical analysis on low-resource CPU-only devices, where existing approaches heavily rely on GPU acceleration. The authors present a CPU-optimized implementation of the MonocularBiomechanics framework through performance-profiling-driven system redesign. Key optimizations include eliminating disk I/O serialization bottlenecks, re-engineering model initialization, and introducing multithreaded parallelism. This is the first demonstration of research-grade, markerless 3D motion analysis operating efficiently in a pure CPU environment. Evaluated on an AMD Ryzen 7 9700X, the optimized pipeline achieves a 2.47× throughput improvement, reduces total runtime by 59.6%, and decreases initialization latency by 4.6×, while maintaining high accuracy—joint angle deviations average only 0.35° (r = 0.998)—thus substantially lowering hardware requirements without compromising analytical precision.

Current markerless Timed Up and Go (TUG) analyses lack robustness and reproducibility, limiting their utility in clinical and research settings. This work proposes tugturn.py, a Python-based, end-to-end markerless 3D TUG analysis pipeline that, for the first time, integrates phase segmentation, gait event detection, and advanced biomechanical metrics—such as Vector Coding and extrapolated Center of Mass (XCoM)—within a unified framework. The method employs spatial thresholds for phase segmentation and a relative distance strategy to identify heel-strike and toe-off events, leveraging 3D pose estimation to generate HTML reports, CSV outputs, and quality-control visualizations. Full reproducibility is ensured through TOML configuration files, while a command-line interface and comprehensive examples enhance accessibility. This pipeline substantially advances the standardization, reliability, and practical applicability of markerless TUG assessment.

This study addresses the high cost and reliance on multi-view setups inherent in traditional optical motion capture systems by proposing a low-cost, monocular video–based approach for quantitative biomechanical analysis in non-laboratory settings. The method introduces the first end-to-end, open-source framework that requires no additional training, seamlessly integrating temporally consistent 4D human mesh reconstruction from SAM-Body4D with the OpenSim biomechanical solver to automatically generate trajectory files compatible with diverse musculoskeletal models. Innovatively, it achieves direct coupling between training-free 4D reconstruction and established biomechanical simulation, supported by an automated prompting strategy and a native Linux processing pipeline. Validation on walking and drop-jump tasks demonstrates knee kinematics prediction accuracy comparable to multi-view systems, highlighting its potential for deployment in home-based environments.

Clinical movement analysis lacks high-quality, markerless biomechanical datasets and general-purpose models tailored for rehabilitation medicine. Method: We introduce BioMotionLM, the first multimodal foundation model for rehabilitation—(1) constructing a 30+ hour, cross-population biomechanical trajectory dataset (including diverse motor disorders) with trajectory tokenization; (2) designing a multimodal Transformer architecture that enables end-to-end alignment between biomechanical trajectories and clinical semantic queries; and (3) releasing a large-scale clinical movement question-answering dataset and performing instruction tuning. Contribution/Results: BioMotionLM significantly outperforms unimodal baselines across five clinically relevant tasks—activity recognition, motor disorder detection, diagnostic inference, clinical scale scoring, and gait quantification—demonstrating strong generalizability, interpretability, and clinical readiness. It establishes a unified foundation model framework for rehabilitation movement analysis.

Current robotic-assisted shoulder rehabilitation lacks real-time biomechanical feedback, compromising tendon safety during therapy. Method: This study introduces the first integration of a high-fidelity OpenSim musculoskeletal model into a real-time robotic closed-loop control framework, enabling online tendon strain estimation and adaptive trajectory replanning. By unifying optimal control, real-time state estimation, and impedance control, the system dynamically models and actively avoids excessive tendon loading across the full glenohumeral range of motion. Results: In healthy subjects, the system successfully executed strain-minimizing trajectories, significantly reducing peak tendon strain—particularly in the supraspinatus—and met clinical real-time requirements (<10 ms control cycle). This work overcomes the limitation of conventional rehabilitation robots that neglect dynamic physiological constraints, establishing a new paradigm for personalized, biomechanics-driven intelligent rehabilitation.

This work addresses the gap between geometric 3D human pose estimation and the biomechanical attributes required in rehabilitation and sports science. We propose BioModule, a lightweight, pose-estimator-agnostic temporal Transformer module that can be appended to any existing 3D pose estimator to predict biomechanically meaningful quantities from standard 17-joint skeletons. To enable frame-level cross-modal supervision, we construct the first large-scale aligned dataset and systematically analyze the impact of upstream pose accuracy on downstream biomechanical prediction performance. By integrating anatomical coordinate alignment with the Human3.6M family of datasets, BioModule demonstrates consistent effectiveness across seven state-of-the-art pose estimators, enabling, for the first time, non-invasive and physically interpretable visual biomechanical analysis.

Existing markerless hand motion capture methods suffer from low accuracy under complex dexterous motions, are highly susceptible to occlusions, and often fail to satisfy biomechanical constraints. To address these limitations, this work proposes an end-to-end, multi-view markerless hand motion capture approach that integrates a differentiable biomechanical hand model into a gradient-based optimization pipeline, jointly refining pose and shape parameters directly from multi-view video inputs. Evaluated on an 8-camera system, the method demonstrates strong robustness to occlusions and intricate hand gestures, achieving a 100% reconstruction success rate across 121 test sequences—significantly outperforming conventional two-stage methods, which attain only 85%. The reconstructed hand motions exhibit superior physiological plausibility, adhering closely to realistic biomechanical behavior.

Existing human-in-the-loop simulation approaches rely heavily on heuristic parameter tuning and lack data-driven personalization, resulting in insufficient fidelity. This work proposes a Real2Sim standardization pipeline that leverages user feedback on “safety and comfort” to identify a 12-dimensional set of individualized parameters at the pelvis-harness interface, using a six-degree-of-freedom viscoelastic model optimized via the CMA-ES algorithm. Intra-class correlation analysis distinguishes universal from subject-specific parameters, while a reproducible operating point eliminates ambiguity in harness tension. Remarkably, only five parameters require calibration to adapt the model to a new user. The calibrated model accurately reproduces real-world interaction envelopes and elicits biomechanically plausible gait adaptations, significantly enhancing simulation fidelity and enabling preclinical validation of personalized controllers.

This study addresses the limited understanding of individual adaptation dynamics during exoskeleton-assisted walking and their temporal evolution. Integrating motion capture, metabolic measurements, and multivariate time-series analysis, the research characterizes dynamic changes in lower-limb kinematics, inter-joint coordination, and metabolic cost at both group and individual levels. The findings reveal that kinematic adaptation exhibits pronounced fluctuations during the swing phase, with asynchronous convergence across joints, while metabolic responses display substantial inter-individual heterogeneity and frequently fail to reach steady state. These results underscore the highly individualized nature of the adaptation process, challenging conventional assumptions of steady-state behavior and highlighting the necessity of modeling individual time-series trajectories to enable precise, personalized exoskeleton assistance.