Applying dynamical component analysis to marker-based gait data: a methodological guide

Abstract In this study, the potential of dynamical component analysis (DyCA) to extract relevant gait dynamics from marker-based whole-body data is investigated. The aim is to outline methodological considerations for the practical application of DyCA for human movement data, while also providing an exemplary use case. Using the public MelbUni2023 dataset and employing a singular value decomposition (SVD)-based DyCA variant, we analyze four selected components individually. Notably, higher modes appear to capture physiologically interpretable nonlinear control pulses—features that often remain obscured in conventional variance-driven methods. The results demonstrate that DyCA isolates main linear dynamics as well as subtle nonlinear movement patterns, presenting the method as a promising tool for future gait analysis in healthy and pathological cohorts.