Motion-Based Sarcopenia Estimation

This research project investigated motion-based and dynamics-based indicators for sarcopenia assessment in older adults. The project focused on estimating ground reaction forces and lower-limb joint torques from human motion data, with the goal of reducing reliance on force plates and enabling more accessible biomechanical assessment during daily-living movements.

The work combined motion capture, OpenSim-based musculoskeletal simulation, ground-reaction-force estimation, and deep-learning sequence models. It produced two related domestic conference presentations: one on ground-reaction-force and joint-torque prediction during squat motion, and another on sarcopenia detection through motion analysis and joint-torque estimation.

• Collected and analyzed motion data from older adults and daily-living movements such as sit-to-stand, step-up, pick-up, and squat.
• Used OpenSim-based musculoskeletal simulation to estimate joint kinematics and joint torques.
• Developed learning-based models for estimating ground reaction forces and ground reaction moments from motion-derived biomechanical features.
• Compared torque-based metrics with established sarcopenia-related indicators such as physical performance, muscle strength, and muscle mass.