Novel Methods for Surface EMG Analysis and Exploration Based on Multi-Modal Gaussian Mixture Models.
Authors: Vögele Anna Magdalena, Zsoldos Rebeka R, Krüger Björn, Licka Theresia
Journal: PloS one
Summary
# Editorial Summary Researchers at the University of Veterinary Medicine Vienna developed a novel statistical approach using Gaussian mixture models (GMMs) to extract meaningful patterns from surface electromyography (sEMG) data collected during equine locomotion, addressing the long-standing challenge of isolating individual muscle activation components from complex, overlapping signals. The team fitted GMMs to sEMG recordings from 14 horses during walk and trot, using four sensor placements (one on the longissimus dorsi and three on the ipsilateral and contralateral gluteus muscles), then applied hierarchical clustering to identify composite peak activation patterns unique to each individual. Key findings revealed that the method successfully isolated distinct activation modes for different muscles within each locomotor cycle, allowing characterisation of muscle-specific firing patterns that would be obscured using conventional peak-and-threshold analysis. For practitioners, this mathematical framework offers a more sophisticated means of interpreting sEMG data—whether for assessing muscular asymmetries, evaluating the effects of therapeutic interventions, or detecting early signs of neuromuscular dysfunction—by distinguishing genuine muscular recruitment patterns from artefactual noise and overlapping activation signals. The hierarchical clustering capability also enables objective comparison of activation patterns between individuals and across different movement phases, supporting more rigorous, quantifiable approaches to equine gait analysis and rehabilitation assessment.
Read the full abstract on PubMed
Practical Takeaways
- •This mathematical approach provides a reproducible method for analyzing muscle activation patterns that could improve understanding of locomotor biomechanics in clinical assessments
- •The technique allows identification of individual variation in muscle recruitment patterns, potentially useful for detecting asymmetries or dysfunction in working horses
- •While primarily a methodological advance, this approach could eventually support diagnosis of movement disorders or evaluation of rehabilitation progress through objective EMG analysis
Key Findings
- •Gaussian mixture models successfully isolated discrete components of muscle activation patterns in surface EMG data from equine locomotion
- •Composite peak models were identified for long back muscles and gluteus muscles across 14 horses during walk and trot
- •Hierarchical clustering of GMM modes enables new approaches to explore and characterize muscle activation patterns during equine movement