Inertial measurement unit sensors with machine‑learned risk scoring can flag a tiny group of Thoroughbreds at much higher risk of fatal musculoskeletal injury, enabling targeted clinical evaluation.
A retrospective study published in the Journal of the American Veterinary Medical Association evaluated whether accelerometer‑based inertial measurement unit (IMU) sensors could identify racehorses at increased risk for fatal musculoskeletal injury (FMI). Investigators analyzed stride data from 28,481 starts by 11,834 Thoroughbreds at 10 US tracks from July 2021 to May 2024.1 Sensors (StrideSAFE) recorded 3‑axis acceleration at 800 Hz from a pocket in the saddle cloth; a machine‑learned algorithm then assigned each run a risk score (RS) from 1 (lowest) to 6 (highest).
Seventy‑four horses sustained an FMI within 120 days of scoring. Horses assigned a highest RS of 6 made up only 122 starts (0.4%) but were 44.6 times more likely to sustain an FMI than horses scored as 1. Approximately 4% (1 in 25) of horses with an RS of 6 incurred an FMI within 120 days. Fetlock fractures accounted for 65% of FMIs in the cohort, and every horse in the highest risk category that later suffered an FMI sustained a fetlock fracture.1
Limitations
Relative vs absolute risk: Although RS‑6 horses had a 44.6× higher relative risk, the absolute risk remained modest (~4% within 120 days). Many flagged horses will not have an FMI (false positives).
Study design: Retrospective observational methods preclude causal inference; confounding and unmeasured management differences may influence outcomes.
Selection bias and generalizability: Tracks that adopt StrideSAFE may differ in surface, training, or management; external validity to other jurisdictions or populations is unproven.
Operational questions: Thresholds for imaging, specific follow‑up protocols, cost, and access to advanced imaging are unresolved.
Associated factors
Multivariable analysis identified several associated factors. Intact males were more likely to receive higher risk scores than females. Shorter race distances were associated with higher RS assignments (the highest‑risk group had a mean race distance of approximately 1372 ± 237 m). The probability of FMI was lower on synthetic all‑weather surfaces compared with dirt and turf; dirt races comprised approximately 55% of starts in the data set.1
Clinical relevance and practical workflow
Authors and the American Veterinary Medical Association frame the IMU‑plus‑algorithm system as a screening or triage tool rather than a diagnostic test.2 The system functions as a “check engine light,” enriching for horses that warrant expedited clinical evaluation, targeted lameness examination, and selective advanced imaging (eg, standing PET or targeted CT/MRI of the fetlock). Because RS‑6 assignments were rare (122 over approximately 145 weeks across 10 tracks, averaging < 1 RS‑6 case per week), concentrating advanced diagnostics on this small group is operationally feasible and could optimize use of limited veterinary resources.1
Bottom line
IMU‑based monitoring with machine‑learned risk scoring can identify a tiny subset of Thoroughbreds at elevated relative risk of catastrophic musculoskeletal injury. Clinicians should interpret algorithmic risk scores alongside clinical exam and imaging, recognize the modest absolute risk among RS‑6 horses, and confirm funding and conflict‑of‑interest disclosures in the published manuscript prior to large‑scale adoption.
References
Mc Sweeney D, Wang Y, Palmer SE, et al. Thoroughbreds deemed to be most at risk by inertial measurement unit sensors suffered a fatal musculoskeletal injury at a higher rate than other racehorses. J Am Vet Med Assoc. Published online September 17, 2025. doi:10.2460/javma.25.04.0268
New technology could help reduce fatal injuries in horse races. News release. American Veterinary Medical Association. November 11, 2025. Accessed November 13, 2025. https://www.avma.org/news/press-releases/new-technology-could-help-reduce-fatal-injuries-horse-races
