Can computer vision “see” the rider’s and surface’s effect on the horse’s symmetry?

Using SLEIP AITM to monitor horses’ locomotor asymmetry

Abstract

Background: Objective assessment of equine locomotor asymmetry is increasingly important for improving lameness detection and monitoring horse welfare. Markerless computer vision systems may provide a practical alternative to traditional gait assessment, although the effects of rider experience and surface type remain unclear.
Objective: To investigate whether a markerless computer vision system (SLEIP AI™) can detect the effects of rider and surface type on locomotor asymmetry and to quantify between-horse variability.
Methods: Four horses were evaluated under six conditions combining rider presence (no rider, novice and experienced rider) and footing type (hard and soft surfaces). Locomotor asymmetry was measured using a smartphone-based system. Linear mixed-effects models assessed the effects of rider and footing, with horse included as a random effect. Intraclass correlation coefficients (ICC) and Fleiss' kappa were used to evaluate variability and asymmetry consistency.
Results: Footing had no significant effect on asymmetry. Rider significantly affected forelimb impact asymmetry (p = 0.030), with greater asymmetry under the novice rider. Between-horse variability explained 36–37% of forelimb asymmetry but only 3–8% of hindlimb asymmetry. Agreement in asymmetry sidedness was moderate overall.
Conclusions: Rider experience influenced locomotor asymmetry more than surface type. Markerless computer vision appears to be a practical tool for objective gait assessment, lameness evaluation and training monitoring.