Multi-Gait Feature Analysis And CoM Inclination Angle Prediction During Rehabilitation Training On Curve Path

In recent years, incidence, morbidity and mortality of various motor nerve systemdiseases have been at an all-time high, such as stroke, spinal cord injury andParkinson’s. Most patients have marching dysfunction, manifested as abnormal gait,difficulty turning, poor flexibility, and not well-done daily activities. Not only thepatients’ own living quality is limited, but also brings a heavy burden on families andsociety. Related studies have confirmed that specific and repetitive walkingrehabilitation training on curve path can remarkably improve patients’ flexibility,coordination and turning ability in walking, compared with walking on straight path.However, gait characteristics, stability characteristics and muscle activation patternsof both lower limbs in curved walking have been less explored and studied, whichdirectly influences the effect and scope of clinical application in motor nervousrehabilitation.To further explore and analyze the proprietary gait characteristics in curvedwalking, this study designed curved walking patterns with three curvatures, threestride lengths and two walking directions, aiming at healthy group (10cases of normalsubjects), model group (8cases of simulating restricted single knee stiffness) andpatient group (6cases of post-stroke hemiplegia subjects). The3-D kinematics dataand surface electromyogram (sEMG) were respectively recorded by VICON motioncapture system and Delsys surface electromyography wireless acquisition system.First, gait cycles were divided according to the minimum vertical coordinate of theheel marker, and, basic gait characteristics under different curved walking patternswere contrasted and analyzed including actual step length, step width, gait cycleduration and velocity. Then, trajectories of CoF (center of feet) and CoM (center ofmass) were obtained based on kinematics data and multiple rigid body model, and,relative radial and tangential inclination of CoM were got to characterize gait stabilityin curved walking. IEMG of8muscles were calculated for the analysis of lowerextremity muscle activation patterns. The results showed that basic gait characteristics,CoM inclination and IEMG during curved walking were significantly different fromstraight walking. This thesis combined8IEMG and immediate CoM radial inclination to predictthe next inclination using support vector regression. Prediction accuracy is0.99, andthe mean square error can be controlled below0.01. In order to improve the useefficiency, muscle-source channels were selected and the prediction scheme wasoptimized according to the correlation coefficients between8IEMG and the CoMradial inclination. Research results in this thesis are expected to provide sometheoretical guidance for making specific rehabilitation training program on curvepath.