Why it matters
Without it, unnoticed lameness or gait irregularities can delay racing plans, risk career-ending injuries, complicate medication control, and threaten regulatory compliance and ownership clarity.
From design to safety
Design
AI design for gait capture and rules
Define data streams from trot-up video, head nod, pelvic asymmetry, stride data, and vet notes; establish decision thresholds and workflow for alerts; document data schemas, retention, audit trails, provenance, and access controls.
- trot-up platform
- high-speed video camera array
- lighting system
- vet notes binder
Build
Integrate capture and storage pipelines
Install 6K60p capable cameras, RGB+IR lighting, fibre links, NVR with RAID 6, and secure cloud hash-backed backups; configure ontology for gait metrics, vet notes, and race entries; formalize calibration cadence.
- high-speed video camera array
- lighting system
- gait analysis workstation
Operation
Daily monitoring and anomaly checks
Capture twice-daily trot-ups, run real-time gait analysis with 120 Hz sampling, generate alerts for asymmetries over thresholds, log events with horse IDs, race entries, and upcoming declarations.
- trot-up platform
- high-speed video camera array
- gait analysis workstation
Maintenance
Data integrity and calibration cycles
Schedule weekly camera calibration using checkerboard targets, lighting stability checks, vet note audits, refresh datasets with last 90 days, validate thresholds quarterly, update access controls and model weightings.
- lighting system
- high-speed video camera array
- vet notes binder
Safety
Welfare and compliance safeguards
Automated flagging reviewed by the yard vet and licensed trainer; ensure withdrawal times, passport, vaccination records, inspections, and licensing align with regulator rules; incident logs maintained.
- vet notes binder
- gait analysis workstation
- portable ultrasound machine
Part by part
- 1
trot-up platform
- Watch for
- Misalignment or uneven surface causing skewed trot analyses.
- AI action
- Adjust camera angle and baseline; trigger retake plan if instability persists.
- 2
high-speed video camera array
- Watch for
- Lighting flicker or lens obstruction impacting frame quality.
- AI action
- Flag poor frames; schedule retake; recalibrate shutter speed and exposure.
- 3
lighting system
- Watch for
- Insufficient illumination or glare reducing feature detectability.
- AI action
- Increase lighting balance; adjust exposure; annotate footage for review.
- 4
vet notes binder
- Watch for
- Missing or outdated veterinary records and withdrawal information.
- AI action
- Flag expired records; prompt upload; correlate with gait anomalies for clinician review.
- 5
gait analysis workstation
- Watch for
- Software drift or data mismatch between systems.
- AI action
- Run integrity check; re-sync datasets; notify supervisor of anomalies.
- 6
portable ultrasound machine
- Watch for
- Inconsistent imaging technique or poor image quality.
- AI action
- Recommend imaging protocol revision; schedule vet assessment; attach findings to horse record.
Worked example
Catching a gait alert from trot-up footage
Illustrative scenario with sample data.
An 8-year-old gelding shows subtle head nod and pelvic shift during routine trot-ups; AI flags potential lameness needing vet review.
Signal in
horse_id: 742, date: 2026-11-05, trot_speed: 3.2 m/s, stride_length: 1.6 m, head_nod_score: 0.28, pelvic_asymmetry: 0.21, stride_variability: 0.07, previous_lameness: none, vet_notes_last: 2026-10-20, entries_due: 2026-11-15, camera_id: CAM-12, lighting_quality: good
System out
alert: gait_anomaly; horse_id: 742; score: moderate; recommended_actions: vet assessment, clinical lameness scoring, consider ultrasound; assign follow-up within 48 hours; rider/exercise plan to be reviewed.
Human sign-off: Yard vet signs off the alert, with the licensed trainer approving the diagnostic plan and schedule for ultrasound if needed; owner notified via secure report.
How to roll it out
Phase 1
Weeks 1 to 2
Baseline capture of all trot-ups across a small group of horses and verify data pipeline; deploy initial alert rules with clinician calibration.
Phase 2
Weeks 3 to 6
Expand to full yard; validate thresholds against clinician scores and adjust workflow; introduce owner-facing summaries in monthly reports.
Phase 3
Weeks 7 to 10
Fine-tune alert routing, reporting cadence, and data retention policies; lock down access controls and audit trails.
Where people stay in charge
- All welfare decisions must be reviewed by the yard vet or licensed trainer before action.
- No withdrawal or treatment decisions can be executed without regulator-compliant authorization.
- Blinded data sharing with owners occurs only via approved reports and with consent.
Questions owners and crews ask
How does AI gait analysis detect lameness from trot-up video?
The system analyzes head nod, pelvic tilt, stride symmetry, and timing; it correlates these with vet notes to flag potential issues for clinician review.
Who reviews gait alerts and decides on treatment?
The yard veterinary surgeon and licensed trainer review alerts; they determine if further imaging or rest is needed before race entries.
What data is stored for each horse?
Video frames, gait metrics, vet notes, withdrawal times, vaccination records, and race-entry statuses are stored with horse IDs.
How do I start integrating this in a busy yard?
Install trot-up platform and cameras, establish data routing to a secure workstation, and train staff to interpret alerts and record vet notes.