Detecting Equine Gaits Through Rider-Worn Accelerometers
Authors: Jorn Schampheleer, Anniek Eerdekens, Wout Joseph, L. Martens, M. Deruyck
Journal: Animals : an Open Access Journal from MDPI
Summary
# Editorial Summary: Detecting Equine Gaits Through Rider-Worn Accelerometers Monitoring equine gait quality is fundamental to training progression and early detection of lameness, yet traditional sensor attachment to horses raises welfare concerns and can alter biomechanics. Schampheleer and colleagues investigated whether accelerometers worn by the rider could reliably classify the four primary gaits (halt, walk, trot, canter) across different body positions—knee, spine, chest, and arm—using data from five riders and seven horses across varying sensor sampling rates and analysis windows. A specialised neural network achieved 89.7% accuracy in gait classification, substantially outperforming seven alternative machine-learning models, with the spine and chest placements proving most reliable for consistent detection. The findings suggest that non-invasive rider-mounted wearable technology could provide practitioners with real-time gait assessment during ridden work, offering coaches and veterinarians an objective tool for monitoring movement quality without compromising horse comfort or natural movement patterns. This approach warrants further validation across diverse populations and ridden disciplines, but demonstrates promising potential as a practical, welfare-friendly addition to equestrian performance monitoring and rehabilitation protocols.
Read the full abstract on PubMed
Practical Takeaways
- •Wearable accelerometers on riders offer a non-invasive method to monitor and assess horse movement patterns during training and rehabilitation
- •Knee-mounted sensors provide the most reliable gait classification, suggesting this placement could be practical for real-world training applications
- •This technology could support objective gait assessment for performance optimization, lameness detection, and welfare monitoring without requiring horse-mounted devices
Key Findings
- •Neural network models achieved 89.7% accuracy in classifying equine gaits (halt, walk, trot, canter) using rider-worn accelerometers
- •Knee sensor placement on riders provided optimal classification performance across different horses and riders
- •Sensor sampling frequency and analysis window length significantly affected classification accuracy
- •Rider-worn accelerometers successfully detected gait transitions without direct attachment to horses, reducing potential discomfort and interference