Gaussian-Laplacian Mixture-Enhanced AGLR for Accurate EMG Onset Detection

Automated detection of muscle contraction onset in electromyography (EMG) signals is crucial for clinical diagnostics and human-machine interaction. However, many existing methods rely on Gaussian assumptions that may not hold for submaximal muscle contractions, where EMG signals display more Laplacian-like characteristics. This study introduces a novel GLM-AGLR (Gaussian-Laplacian mixture model based approximated generalized likelihood ratio) algorithm to address this limitation. By incorporating a Gaussian-Laplacian mixture model for the contraction phase, GLM-AGLR achieves more accurate onset detection compared to conventional approaches. Experiments on EMG signals collected at various contraction intensities reveal that GLM-AGLR consistently outperforms both threshold-based methods and earlier AGLR variants, yielding results closer to visually identified ground truth. These findings suggest that the proposed model not only bridges the gap between Gaussian and Laplacian assumptions but also enhances the reliability of EMG signal segmentation and feature extraction, ultimately supporting more robust applications of EMG-based technologiesClinical relevance- The proposed algorithm improves automated muscle contraction onset detection, providing insights towards improving clinical applications, such as muscle diagnostics or EMG-based technologies for patients.