Research On Human Motion Intention Analysis And Rehabilitation Assessment System For Lower Limb Rehabilitation

Objective: The seventh census,China’s population aged 60 and above exceeds 260 million people,accounting for 18.87% of the total population,and the aging of the population has intensified.Stroke has become one of the biggest pathological factors endangering the health and quality of life of people in China,and about 80% of stroke patients have sequelae,of which 54% are with lower limb walking impairment,and the form of disability assistance is severe.The aim of this study is to investigate the modular control strategy of human lower limb walking motion,and construct a continuous estimation model of lower limb multi-joint angle.And further study the differences between stroke patients and healthy adults,quantify and integrate kinematic and electrophysiological parameters,and establish a human motion intention analysis and multimodal collaborative quantitative rehabilitation assessment system for lower limb exoskeleton to improve the effectiveness and efficiency of clinical rehabilitation.Methods: Based on muscle synergy theory,this study analyzed lower limb movements through Opensim simulation and experiments to reveal the modular neural control strategy of lower limb walking.By collecting 8-channel EMG signals from 8 healthy adults with uniform step speed of 2km/h,3km/h and 5km/h in the lower limbs,the effects of different step speeds and different EMG characteristics on the results were compared and analyzed,and a continuous estimation model of multi-joint angular co-activation in the lower limbs was established.To help patients walk on the ground as early as possible,a rehabilitation assessment technique for the lower limb exoskeleton rehabilitation robot was developed.EMG and kinematic data were collected from 10 patients with different Brunnstorm stages,and the similarity of the synergistic features of EMG and kinematics with the established baseline template was calculated using a synergistic quantification algorithm to compare the correlation between the synergistic features and clinical scales according to the lower limb motion characteristics.The information at the electrophysiological and kinematic levels was fused to build a modal feature fusion model to output lower extremity motor function scores,and several commonly used machine learning methods were validated.Results: The experimental results showed that the timing and composition of each of the four basic control modules of lower limb walking decomposed by the present study method were robustly consistent across subjects,and each module corresponded to the early support,late support,early swing,and late swing of the gait cycle,respectively,and were consistent with the biomechanical task of walking.The continuous estimation model of lower limb multi-joint angular co-activation proposed in this study can effectively analyze the subjects’ motor intention,in which the closer the gait speed is to that of a healthy adult,the better the prediction is,and the coactivation characteristics also show better prediction and robustness compared to the traditional EMG characteristics,in which the best regression coefficients of 0.77,0.75,0.68 are obtained at 3km/h lower limb hip-knee-ankle joint,respectively.For the rehabilitation assessment technique of the lower limb exoskeleton rehabilitation robot,the experimental results showed that the proposed method obtained higher correlation coefficients with the clinical scale in the K-nearest neighbor(KNN)model(R=0.921,P<0.001),which can modify the rehabilitation training mode of the exoskeleton robot according to the assessment index,and the results have good generalization and can be applied to the actual clinical scenarios of rehabilitation assessment;in addition In addition,the EMG sensors and IMU sensors required for assessment are simple to use and easy to wear,which can realize the "human-in-the-loop" assessment and training synchronization mode on the basis of the lower limb exoskeleton without additional sensors.Conclusion: The lower extremity rehabilitation assessment system based on human motion intent analysis proposed in this study provides personalized rehabilitation protocols and recommendations to help subjects improve their rehabilitation effectiveness and efficiency.In addition,the system can monitor the progress of the rehabilitation process and make adjustments and optimization of the rehabilitation program.It can provide powerful support to rehabilitation doctors and patients to help them better develop rehabilitation plans,monitor rehabilitation progress and improve rehabilitation outcomes.In the future,the system can be further optimized and improved to expand its application in the field of rehabilitation.