Machine Learning for Personalisation of Biomechanical Movement Simulations (C01)

The aim of this subproject is to investigate machine learning methods for the personalization of bio-mechanical models for individual movement analysis and prediction (synthesis) of movement changes. Methods are developed to automatically personalize musculoskeletal models and movement target functions based on heterogeneous empathokinesthetic measurement data and to extract individual movement patterns. The acquired knowledge can be used to explore individual differences in the musculoskeletal system and movement control, as well as to predict movement changes when no or hardly any measurement data are available. Individualized movement analyses and predictions thus provide information for intervention planning (e.g. of prostheses or operations) or enable an objective diagnosis of changes in the gait pattern.