AI-assisted digital video analysis reveals changes in gait among three-day event horses during competition.
Authors: Bucci Madelyn P, Dewberry L Savannah, Staiger Elizabeth A, Allen Kyle, Brooks Samantha A
Journal: Journal of equine veterinary science
Summary
# Editorial Summary Objective assessment of equine gait remains challenging in field settings, yet understanding how locomotor patterns change during competition has important implications for performance and welfare in three-day eventing. Researchers deployed artificial intelligence-assisted video analysis to track gait changes in 194 international-level event horses across five venues, capturing high-resolution footage at trotting inspections before and after the cross-country phase using consumer-grade cameras and the DeepLabCut software to identify 26 anatomical keypoints per frame. Significant alterations emerged in duty factor (the proportion of stride cycle spent in contact), forward speed, and forelimb swing range post-cross-country (P ≤ 0.05), demonstrating measurable biomechanical effects of competition exertion on locomotion. The AI-driven approach processed all 388 videos in minutes rather than the months manual analysis would require, whilst eliminating the need for expensive motion-capture systems or force plate equipment. For practitioners seeking objective, repeatable gait assessment—whether for post-competition evaluation, rehabilitation monitoring, or research—this method offers a scalable solution that could become standard for quantifying performance-related changes in working horses.
Read the full abstract on PubMed
Practical Takeaways
- •AI video analysis offers a practical, affordable tool for objective gait assessment at competitions without specialized equipment, enabling evidence-based evaluation of horse fitness and welfare
- •Measurable gait changes (duty factor, speed, forelimb swing) following cross-country work provide quantifiable indicators of fatigue and exertion effects that could inform competition safety and recovery protocols
- •This technology could standardize gait evaluation across venues and riders, moving away from subjective assessments toward objective data for veterinary decision-making and performance management
Key Findings
- •AI-based DeepLabCut analysis successfully quantified gait in 388 videos in minutes versus months of manual labeling
- •Significant changes in duty factor, speed, and forelimb swing range detected post-cross-country phase (P ≤ 0.05)
- •Consumer-level video camera with AI analysis provides economical, objective, and repeatable field assessment of equine locomotion
- •Twenty-six keypoint tracking method enables quantitative gait parameter derivation suitable for large-scale biomechanical studies