End-effector robots for gait rehabilitation allow an intensive, controlled and repetitive training of the lower limbs. These devices assist the gait cycle by providing body weight support while applying mechanical forces directly to the user’s feet. Several studies have shown how end-effector robots can improve gait kinematics, however little research has been conducted to investigate how the device might affect the gait muscle activation patterns of the user. Our aim was to investigate: (i) if the device induces different muscle patterns than those adopted during overground walking and (ii) whether and how different levels of assistance influence muscle patterns while walking with the device. 16 healthy volunteers walked 100 steps overground and with the support of an end-effector robotic device (GEO-system). In the latter case, participants were asked to walk while the device provided maximum, adaptive, and no bilateral assistance. During walking we recorded, using an electromyographic system (EMG), the activity of 8 muscles from each leg: tibialis anterior (TA), gastrocnemius medialis (GM), soleus (SOL), rectus femoris (RF), vastus medialis (VM), semitendinosus (ST), biceps femoris (BF), gluteus medius (GLM). At the same time, 2 Inertial Measurement Units (IMUs) were used to measure the angular velocity and angles of both feet. The gait cycle events were detected using the IMUs data and were used to segment the EMG signals. The latter were used to extract, with a non-negative matrix factorization algorithm [1], the muscle synergies - weights coefficients and activation profiles - of each walking condition. The same paradigm was also adopted to investigate the changes in muscle strategies during the ascent - descent of stairs. In the following, we will only describe and discuss the results of the overground walking task. Differences in the muscle activation patterns were observed between overground and end-effector assisted walking. When walking with the device, regardless of the assistance provided, the activity of the distal muscles (TA, SOL, GM) decreased, as well as the activity of most of the proximal muscles (GLM, ST, BF). No relevant differences were observed in muscle activations of the knee extensors (RF, VM), while changes in timing were observed in the muscle activations of the TAs, STs and BFs. In the end-effector assisted walking, the TAs were active only in the swing phase of the gait cycle, while the STs and BFs were active only in the stance phase of the gait cycle. Moreover, no differences in the muscle activation timings were observed when walking with the device in all assistive modalities. The only changes induced by the type of assistance were at the level of the STs and TAs. The activity of both muscles in the no assistance condition was higher than the maximum and adaptive assistance conditions. The 4 muscle synergies extracted for each walking condition confirmed these observations. In synergy 1, which is mainly described by ST, BF and TA, all the end-effector walking conditions appear to have a lower contribution of the ST and TA. In synergy 2 and 3, described respectively by the TA and the GLM, the weights during end-effector assisted walking are the same as those of the overground walking. In synergy 4, which is mainly described by the SOL and GM, the weights of both muscles in the end-effector walking conditions display a smaller contribution than in the overground walking condition. As expected, the activation profile of the muscle synergies in the end-effector and overground walking conditions appear to have different activations timings (see Table 1). The differences observed in the muscle patterns between end-effector assisted walking and overground walking appeared more influenced by the mechanical structure of the device than by the assistance provided. Specifically, the mechanical constrains due to anchoring the feet of the user to the platforms of the device and to the absence of a forward progression in space while walking may be the reasons behind the muscle patterns differences.
Read full abstract
May 14, 2024
Paul Diego + 6
Open Access
