Research On Limb Rehabilitation System Based On Deep Learning And Multimodal Sensor Dat

Limb dysfunction refers to the clinical lesions of the limbs which are not controlled by mind.Effective rehabilitation training and muscle fatigue monitoring are basic treatment programs for patients with limb dysfunction,aiming at avoiding disuse atrophy of limb muscles and improving the immunity of the organism.At present,medical resources in our country are limited,and rehabilitation costs are high.There are a large number of patients with limb dysfunction,but the traditional rehabilitation training mode is inefficient.The comprehensive limb rehabilitation training evaluation system is deficient in China.With the proposal of intelligent rehabilitation,the key to tracking and managing patients’ rehabilitation lies in the efficient and accurate identification and acquisition of patients’ rehabilitation training movements,and realtime muscle fatigue evaluation of patients’ rehabilitation is needed to ensure the safety of rehabilitation training.In view of the current needs of patients with limb dysfunction,this study obtains the rehabilitation data of patients through intelligent perception technology with the safe,real-time,non-invasive and intelligent rehabilitation assessment as the goal and patients with limb dysfunction as the research object,and makes full use of the advantages of the data itself.This study is committed to the realization of rehabilitation training and muscle fatigue assessment of patients with limb dysfunction.The main contents and achievements of this thesis are as follows:(1)In this study,a light and convenient multi-mode sensor is adopted to acquire the movement data of patients during rehabilitation training.During the training,patients can avoid limb movement injuries caused by carrying heavy objects.Given the limited medical resources and the impact of the COVID-19 epidemic,it is more difficult for patients to travel to and from hospitals.This study developes a rehabilitation training platform based on multi-mode sensor technology,so that patients can carry out highquality rehabilitation training at home according to the rehabilitation plan formulated by doctors.The rehabilitation training platform sends the rehabilitation training data of patients to the server side through Internet technology,and the server side will input the rehabilitation training data into the established limb function evaluation model after receiving it.Taking elbow flexion commonly seen in upper limb rehabilitation as an example,this study constructes an evaluation model based on dynamic time planningK Nearest Neighbor(DTW-KNN)algorithm and an evaluation model based on long short-term memory(LSTM)neural network to evaluate limb function.The limb function evaluation model can objectively and effectively evaluate the motor function level of patients in the process of rehabilitation training in real time,and can reduce the psychological obstacles of rehabilitation and improve the enthusiasm of patients in rehabilitation training.(2)Studies have found that patients with limb dysfunction have low muscle sensitivity,and there is a large deviation between subjective fatigue perception and actual muscle function.Therefore,in order to recover soon,patients tend to ignore muscle fatigue and carry out high-intensity and long-term rehabilitation training,which will result in negative effects.In this study,the surface electromyography(s EMG)sensing equipment is used to obtain the electromyography data of patients during rehabilitation training,and the advantages of s EMG’s noninvasive,real-time and multitarget measurement are utilized to accurately evaluate the muscle fatigue status of patients in real time.Compared with previous studies,which only focused on a certain stage of muscle fatigue detection,so the overall detection of muscle fatigue state is inadequate.In this study,by capturing the electromyography data of the whole powerrelaxation stage of rehabilitation movements,high-quality evaluation of patients’ fatigue state is achieved.In this study,the obtained s EMG data is input into the constructed muscle fatigue evaluation model,and the experimental results show that the Attention-LSTM algorithm achieves higher accuracy than the LSTM algorithm.Therefore,the muscle fatigue evaluation method based on attention mechanism and s EMG proposed in this study can objectively quantify the degree of muscle fatigue of patients and provide safety guarantee for rehabilitation training.