Upper Limb Joint Angle Prediction Method Based On Lssvm

In recent years, the incidence of stroke has been increasing. In order to enhance patient engagement and improve rehabilitation outcomes in active training, it is necessary to obtain real-time information about the patient's movement intention. This paper addresses the issues of poor accuracy and slow prediction speed in current models for predicting patient movement intention. A method based on Least Squares Support Vector Machine (LSSVM) is proposed to predict upper limb joint angles. By continuously estimating the joint angles based on surface electromyographic signals acquired from normal upper limbs, these estimated angles serve as control guidance for the rehabilitation robot of the affected upper limb. In this paper, the patient's movement intention is defined as continuous joint angles. Surface electromyographic signals are obtained as input features, while the patient's joint angles serve as output features. LSSVM is employed as a regression method to predict the trend of joint angle changes, aiming to predict the patient's movement intention.