Research on Prediction Model of Electromyography Signals Based on Multi-Factor Fusion and Spatiotemporal Graph Convolutional Network

’The task of human motion state prediction based on biological information has become a hot topic in biomedical engineering research field. However, due to the complexity of model construction, the lack of prediction accuracy and the interference of external factors, the development of the technology is hindered. Therefore, this paper presents a method to transform the human motion state prediction task into a mapping function optimization problem, considering both time and space factors. This method combines the advantages of graph convolutional network and gated cycle unit to construct a prediction model with integrated spatiotemporal features. The attribute feature fusion module is introduced to integrate external factors such as fatigue and emotional state into the model to enhance the robustness of the model. The results show that the proposed model integrates the spatiotemporal dependence, further improves the accuracy of prediction results, and maintains a stable performance in long-term EMG prediction tasks. The results not only provide a new method for the recognition of human movement signals, but also provide effective guidance for the study of movement rehabilitation control.