Real-Time Knee Angle Prediction Using EMG and Kinematic Data with an Attention-Based CNN-LSTM Network and Transfer Learning Across Multiple Datasets

2025-10-15

AI Summary

To address challenges in real-time knee angle prediction for rehabilitation—namely poor latency, non-representative testing conditions, and scarcity of labeled data—this paper proposes a lightweight attention-enhanced CNN-LSTM model and a multi-center transfer learning framework. The model fuses surface electromyography (EMG), historical joint angles, and human–exoskeleton interaction forces: CNNs extract local spatial features, LSTMs capture temporal dependencies, and an attention mechanism dynamically weights critical time steps. With only a few gait cycles from a new subject, the model rapidly adapts, drastically reducing annotation effort. Evaluated on the SMLE exoskeleton platform, it achieves normalized mean absolute errors (NMAE) of 1.09% (1-step) and 3.1% (50-step) under multimodal input, demonstrating robustness to atypical gait patterns. The approach delivers high accuracy, low inference latency (<10 ms), and strong cross-subject and cross-center generalization.

Technology Category

• Intelligent Robots: State Estimation
• Machine Learning: Multimodal Learning
• Natural Language Processing: Safety and Robustness

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Large pretrained models with web data
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Electromyography (EMG) signals are widely used for predicting body joint angles through machine learning (ML) and deep learning (DL) methods. However, these approaches often face challenges such as limited real-time applicability, non-representative test conditions, and the need for large datasets to achieve optimal performance. This paper presents a transfer-learning framework for knee joint angle prediction that requires only a few gait cycles from new subjects. Three datasets - Georgia Tech, the University of California Irvine (UCI), and the Sharif Mechatronic Lab Exoskeleton (SMLE) - containing four EMG channels relevant to knee motion were utilized. A lightweight attention-based CNN-LSTM model was developed and pre-trained on the Georgia Tech dataset, then transferred to the UCI and SMLE datasets. The proposed model achieved Normalized Mean Absolute Errors (NMAE) of 6.8 percent and 13.7 percent for one-step and 50-step predictions on abnormal subjects using EMG inputs alone. Incorporating historical knee angles reduced the NMAE to 3.1 percent and 3.5 percent for normal subjects, and to 2.8 percent and 7.5 percent for abnormal subjects. When further adapted to the SMLE exoskeleton with EMG, kinematic, and interaction force inputs, the model achieved 1.09 percent and 3.1 percent NMAE for one- and 50-step predictions, respectively. These results demonstrate robust performance and strong generalization for both short- and long-term rehabilitation scenarios.

Problem

• Predicting knee joint angles in real-time using EMG signals
• Overcoming limited datasets and non-representative test conditions
• Enhancing rehabilitation with transfer learning across multiple datasets

Innovation

• Attention-based CNN-LSTM model for knee angle prediction
• Transfer learning across multiple datasets for generalization
• Integrates EMG, kinematics, and force data for accuracy