Author(s): WALLNOEFER, E., KOLLER, W., CSAPO, R., BACA, A., KAINZ, H., Institution: UNIVERSITY OF VIENNA, Country: AUSTRIA, Abstract-ID: 2411

INTRODUCTION:
Muscle architecture and coordination adapt to sport-specific demands and movement habits. Consequently, a single exercise modification (e.g., foot position or knee angle during calf raises) may redistribute force among synergists differently across individuals and athletic groups. These relationships remain poorly understood, yet they are critical for targeted training, rehabilitation, and overuse prevention. Musculoskeletal simulations can estimate muscle forces, but generic models - derived from non-athlete data and optimized for activation minimization - may miss personal strategies. We evaluated electromyography(EMG)- and ultrasound(US)-informed workflows personalized to one athlete and quantified how modeling choices alter the triceps surae (TS) force share during in- and out-toeing plantarflexions.
METHODS:
One competitive female Latin dancer performed plantarflexions at varied intensities and foot positions on a dynamometer and walking trials. We recorded 3D motion, EMG, dynamic 2D US of fascicle behavior, and synchronized ankle torque and ground reaction forces. Muscle morphology was measured via 3D freehand US. We built and calibrated multiple models with different levels of personalization, movement combinations, and optimization settings in OpenSim. 6426 workflows were screened for physiological plausibility (fiber lengths, low activation during passive movement, low residuals) and ranked using normalized RMSE and cross-correlations against joint moments, EMG, and fascicle length. A regression model assessed which tools most improved performance. Primary outcomes compared a common generic with the best-performing workflow on TS force share during in- and out-toing plantarflexions.
RESULTS:
The generic workflow estimated gastrocnemius lateralis–medialis–soleus force shares of 10–51–39% (out-toeing) and 13–35–52% (in-toeing). The best-performing workflow estimated 11–34–55% (out-toeing) and 20–28–52% (in-toeing). The top 30 workflows consistently included personalized maximum isometric forces for TS muscles and Achilles tendon moment arm from 3D freehand US, most were calibrated with a combination of a high- and low-intensity trial and balanced inclusion of EMG and US data but rarely integrated US data in the simulations that were run after model calibration. In total, 985 workflows outperformed the generic approach.
CONCLUSION:
Personalizing key parameters (muscle strength, tendon moment arm) and calibrating across mixed intensities appear crucial among our tested workflows. The generic simulation showed increased soleus share for in-toing and aligned poorly with EMG and US data in this athlete. These results opposed the shift between the gastrocnemius heads estimated by the highly personalized approach. Although confirmation in larger samples is needed, it seems that personalized workflows appear well suited for investigating individual responses to simple exercise modifications and would support a more targeted training, rehabilitation, and overuse prevention.