Why it matters
Without AI review, airway issues can be missed or misinterpreted, leading to hidden performance limitations, medication mismanagement, or regulatory concerns during stable inspections and raceday entries.
From design to safety
Design
AI-enabled airway assessment scope
Define data streams from endoscope videos, larynx motion, soft palate dynamics, mucus scoring, and tracheal wash results to inform training plans, welfare targets, and baseline health benchmarks across the yard. Establish data governance, surgeon-friendly metrics, and a consent-ready audit trail for every horse, with versioned baselines and a master dataset linking to passport IDs and trainer notes.
- endoscope videos
- larynx
- soft palate
- mucus scores
Build
Systems integration and validation
Set up video capture rigs at 1080p or higher, 60fps where feasible, standardize mucus scoring on a 0–3 scale with blinded expert panels, ingest tracheal wash data including cytology and bacterial culture results, and calibrate AI thresholds against two independent expert reviews in a pilot cohort of 12 horses with double-blind annotations.
- endoscope videos
- larynx
- soft palate
- tracheal wash
Operation
Real-time airway monitoring
AI processes live or recorded endoscopy, highlights oddities in airway motion, mucus load, and wash results for immediate flags and follow-up scheduling with the vet team; generate daily airway health summaries for yard manager and owners’ portals, with raceday contingency notes.
- endoscope videos
- larynx
- mucus scores
- tracheal wash
Maintenance
Quality checks and retraining
Regularly update AI with new expert annotations, verify endoscope calibration against phantom targets, and review mucus scoring consistency across operators and sessions; run monthly drift checks and track annotation concordance to keep false positives low.
- endoscope videos
- larynx
- soft palate
- tracheal wash
Safety
Regulatory and welfare guardrails
Ensure findings are reviewed by the yard veterinarian and signed off for medication decisions, workload changes, raceday declarations, and welfare alerts that trigger welfare audits; document regulator-facing reports with clear audit trails and responsible officer sign-offs.
- endoscope videos
- larynx
- soft palate
- tracheal wash
Part by part
- 1
endoscope videos
- Watch for
- Video feed artifacts, partial frame loss, inconsistent lighting, or sudden motion blur indicating poor capture.
- AI action
- AI detects frame quality issues and switches to secondary camera angle or requests fresh capture.
- 2
larynx
- Watch for
- Unilateral arytenoid movement, flickering cords, or irregular abduction during phonation.
- AI action
- AI flags potential laryngeal dysfunction and earmarks for expert review.
- 3
soft palate
- Watch for
- Soft palate flutter, arytenoid-tongue interference, or obstructive motion during inspiration.
- AI action
- AI identifies aberrant soft palate dynamics and prioritizes scope review.
- 4
mucus scores
- Watch for
- Elevated mucus scoring thresholds or inconsistent scoring among observers.
- AI action
- AI harmonizes scores across sessions and triggers clinical evaluation if thresholds exceeded.
- 5
tracheal wash
- Watch for
- Contamination, inadequate sample, or unusual cytology.
- AI action
- AI flags suboptimal samples and prompts repeat wash and lab QC.
- 6
endoscopy processing workstation
- Watch for
- Data sync failures, mislabeled files, or missing patient identifiers.
- AI action
- AI flags data integrity issues and prompts metadata correction.
Worked example
Catching a breath test anomaly before a race
Illustrative scenario with sample data.
A yard uses AI-assisted endoscopy to assess a flat racehorse with suspected airway noise during gallops.
Signal in
endoscope_video_id=V321, horse_id=H98, date=2026-10-12, larynx_score=2, soft_palate_motion=moderate, mucus_score=1, tracheal_wash_result=normal, rider_id=R7
System out
AI alert: airway motion irregularity detected in larynx with increased soft palate flutter; mucus score rising to 2; recommend repeat endoscopy and veterinary review.
Human sign-off: Yard veterinarian reviews the AI alert alongside the laryngeal exam and confirms plan with the licensed trainer before any race entry or medication decision.
How to roll it out
Phase 1
Weeks 1 to 4
Baseline data collection from 8 horses and initial expert annotations for AI calibration.
Phase 2
Weeks 5 to 8
System tuning, threshold setting, and begin pilot reporting with welfare flags.
Phase 3
Weeks 9 to 12
Full integration into entry and raceday workflow with documented sign-offs.
Where people stay in charge
- All AI-driven decisions on withdrawal times, medication and raceday entries must be reviewed and signed by the yard veterinarian and the licensed trainer.
- AI outputs that conflict with regulatory guidelines are escalated to the regulator liaison and paused until clearance.
- Data privacy and patient identifiers must be maintained; any sharing requires owner consent and steward approval.
Questions owners and crews ask
How does AI assist endoscopy in a race yard workflow?
AI processes videos and scores from endoscopy, larynx, soft palate, and mucus results to flag abnormalities; it suggests review steps and helps schedule follow-up tests; it integrates with patient identifiers and session metadata.
Who signs off on AI-identified airway findings for race entries?
The yard veterinarian signs off airway findings and the licensed trainer approves any raceday entries or withdrawal decisions based on AI-supported reports, with a written audit trail.
What documentation is generated by AI for airway care?
Structured airway review sheets, mucus scoring trends, tracheal wash results, endoscopy quality checks, and a sign-off log for regulatory compliance; includes data lineage and version history.
How is data quality ensured in AI airway monitoring?
Regular calibration with expert reviews, standardized capture protocols, automated data integrity checks, and cross-validation against independent assessor annotations to prevent drift.