Study finds markerless equine gait analysis aligns with IMU systems: full analysis
A new wave of evidence is strengthening the case for markerless gait analysis in horses. In a study featured in Equine Veterinary Journal’s July 24, 2024, “On the Hoof” podcast, researchers reported that objective movement asymmetry measurements from a markerless AI video system were comparable with those from two established sensor-based systems in horses assessed under field conditions. The paper was first published online April 2, 2024, and later appeared in the January 2025 print issue of the journal. (evj.podbean.com)
The question behind the study is a practical one for equine practice: can a smartphone-based, markerless system deliver clinically useful information without the time, hardware, and handling demands of body-mounted sensors? Objective gait analysis has been used for years to support lameness workups, but uptake can be limited by cost, setup complexity, and workflow friction. Markerless systems aim to lower those barriers by using computer vision to extract movement data from ordinary video, potentially making repeat assessments easier in ambulatory and field settings. (pmc.ncbi.nlm.nih.gov)
In the Kallerud study, investigators collected straight-line trot data from 52 client-owned horses in regular training, with 41 horses included in the final analysis. They compared a markerless AI system, Sleip AI, recorded on a tripod-mounted iPhone 14 Pro, against two inertial sensor systems, Equinosis Q with Lameness Locator and EquiMoves, along with subjective evaluation. The main finding was moderate agreement among the objective systems in classifying asymmetric limbs: kappa was 0.59 for forelimbs and 0.44 for hindlimbs. The authors concluded that the systems were comparable under field conditions when defined asymmetry thresholds were applied, and that many discrepancies were driven more by threshold effects than by disagreement on which side was asymmetric. (pmc.ncbi.nlm.nih.gov)
There were, however, important caveats. EquiMoves analyzed significantly more strides than the other systems, and Lameness Locator analyzed more hindlimb strides than the markerless system. The authors specifically noted that Sleip AI analyzed significantly fewer hindlimb strides than the IMU systems, which could be a limitation in clinical use. That nuance matters, because hindlimb lameness is often among the more difficult areas of equine gait assessment, and a tool that performs well overall may still need careful interpretation in those cases. (pmc.ncbi.nlm.nih.gov)
The broader industry context suggests why this paper is getting attention. Markerless and AI-assisted gait tools are moving quickly from research settings toward everyday use, with reviews describing them as part of a broader shift to field-deployable, user-friendly objective gait analysis. Commercial momentum is building too: in July 2024, a report on a Boehringer Ingelheim-Sleip partnership framed markerless AI as a way to expand access to lameness detection and management tools for the equine veterinary community. That doesn't validate any one product on its own, but it does show that industry sees practical demand for lower-friction objective assessment. (mdpi.com)
Why it matters: For veterinary professionals, the study supports a more pragmatic view of objective gait analysis. Markerless systems may help practices gather repeatable asymmetry data with less setup burden, which could be useful for baseline exams, rechecks, prepurchase discussions, treatment monitoring, and communication with pet parents. But the study also reinforces that objective tools are not interchangeable in every circumstance, and that threshold settings, stride capture, and case selection can all shape interpretation. In other words, the technology may improve consistency and access, but it still sits within, not outside, the clinical exam. (pmc.ncbi.nlm.nih.gov)
That point is especially relevant as equine medicine absorbs more AI-enabled tools. A recent commentary in the Journal of Veterinary Internal Medicine called for stronger AI literacy in veterinary medicine, reflecting a wider need to understand not just what these systems can do, but where their blind spots are. For equine clinicians, the opportunity is clear: markerless systems could make objective gait analysis easier to use more often. The responsibility is equally clear: know the limitations, validate findings against the horse in front of you, and use the data to sharpen, not substitute for, clinical judgment. (academic.oup.com)
What to watch: The next phase will likely focus on validation in more diverse clinical populations, refinement of hindlimb performance, and wider use of longitudinal tracking features that may help veterinarians monitor subtle changes over time rather than relying on single-visit snapshots. (sleip.com)