Common gait modulation by presence of hip assistance, individual motor strategies by assistance timing

Hip-assist soft exosuits are tuned by shifting force-profile timing, yet it remains unclear whether responses to different force profiles generalize across individuals or are largely person-specific.

In 22 healthy adults walking at 1.1 m/s (two n = 11 cohorts), we compared a suited baseline (Assist-off) with eight timing-varied hip assist force profiles (A-H) parameterized by onset-peak-offset over the gait cycle, and evaluated prolonged exposure to a single profile. Lower-limb kinematics and electromyography (EMG) were measured. Cohort-level differences were tested using repeated-measures statistics with false discovery rate correction. Stride-wise separability for kinematics and EMG was evaluated using linear discriminant analysis (LDA) with leave-one-subject-out (LOSO) and within-subject cross-validation.

Across force profiles, hip assistance induced a consistent response relative to Assist-off: increased peak hip flexion (7 of 8 profiles), increased peak ankle dorsiflexion (8 of 8 profiles), and reduced soleus activation (6 of 8 profiles), while average differences among profiles remained small. During prolonged exposure to a single profile, these overall assist-induced responses were further accentuated over time. Stride-level classification of kinematic and EMG features showed weak cross-participant generalization but strong within-participant separability (macro-F1 > 0.98), indicating individualized force-profile-dependent signatures.

Hip exosuit responses suggest a two-layer organization-an assistance-presence-driven, profile-invariant cohort-level response and individualized force-profile-dependent signatures-indicating that evaluation and optimization of wearable assistance should account for both shared and person-specific responses to force-profile variation.