Synergy-Based Estimation of Balance Condition During Walking Tests

In the area of human-machine interface research, the continuous estimation of the Center of Pressure (COP) in the human body can assess users’ balance conditions, thereby effectively enhancing the safety and diversity of studies. This paper aims to present a novel method for continuous synergy-based estimation of human balance states during walking, and simultaneously analyze the impact of various factors on the estimation results. Specifically, we introduce muscle synergy coherence features and analyze the variations of these features in different balance conditions. Furthermore, we fuse temporal features extracted by a bidirectional long short-term memory (BILSTM) network with spatial features derived from the analysis of muscle synergy coherence to continuously estimate the mediolateral COP and Ground Reaction Force (GRF) during human walking tests. Then, we analyze the influence of different electromechanical delay compensation (EMD) time, the number of synergies, and different walking speeds on the estimation results. Finally, we validate the estimation capability of the proposed method on data collected in real-world walking tests. The results indicate a significant correlation between the proposed muscle synergy coherence features and balance conditions. The network structure combining muscle synergy coherence features and BILSTM features enables accurate continuous estimation of COP ( $\mathbf {R}^{\mathbf {{2}}}= \,\, 0.87~\pm ~0.07$ ) and GRF ( $\mathbf {R}^{\mathbf {{2}}}= \,\, 0.83~\pm ~0.09$ ) during walking tests. Our research introduces a novel approach to the continuous estimation of balance conditions in human walking, with potential implications in various applications within human-machine engineering, such as exoskeletons and prosthetics.