Inertial Sensor-Based Upper Limb Rehabilitation Auxiliary Equipment and Upper Limb Functional Rehabilitation Evaluation

Abstract

AbstractStroke patients in rehabilitation period often encounter problems such as high training cost and weak self supervision. We proposed a rehabilitation evaluation method of upper limb function based on wearable inertial sensor data acquisition auxiliary equipmen, which aims to realize the self-monitoring and evaluation of rehabilitation of patients in the middle and late stage of upper limb function rehabilitation. We have used three inertial sensing units MPU6050 to make wearable upper limb rehabilitation training auxiliary equipment, which can collect the motion data of stroke patients in rehabilitation period during daily rehabilitation training. Combined with lindmark upper limb rehabilitation scale, we collected eight hand gesture data for upper limb rehabilitation exercise evaluation. For the original data, quaternion data and Euler angle data, we established upper limb rehabilitation training action evaluation models based on libsvm, multi-layer LSTM and cnn-lstm neural network respectively to evaluate the rehabilitation status of patients. The results show that CNN LSTM model has the best performance, with the recognition accuracy of 99.67%, followed by multi-layer LSTM, and the model recognition accuracy of 97.00%. The work of this paper will provide a reference for patients in the middle and later rehabilitation stage of upper limb after stroke to realize their own supervised rehabilitation training and recovery state evaluation.KeywordsInertial sensorsWearable devicesUpper limb rehabilitation training and assessment