FREE SHIPPING over $40
OVER 9000 FIVE-STAR REVIEWS
5% OFF ALL SUBSCRIPTIONS
100% HAPPINESS GUARANTEE
Analyze Diet
0
Home / Equine Research Bank / Sci Rep / A machine learning approach to identify stride characteristi...
Scientific reports2024; 14(1); 28967; doi: 10.1038/s41598-024-79071-1

A machine learning approach to identify stride characteristics predictive of musculoskeletal injury, enforced rest and retirement in Thoroughbred racehorses.

Abstract: Decreasing speed and stride length over successive races have been shown to be associated with musculoskeletal injury (MSI) in racehorses, demonstrating the potential for early detection of MSI through longitudinal monitoring of changes in stride characteristics. A machine learning (ML) approach for early detection of MSI, enforced rest, and retirement events using this same horse-level, race-level, and stride characteristic data across all race sectionals was investigated. A CatBoost model using features from the two races prior to an event had the highest classification performance (sensitivity score for MSI, enforced rest and retirement equal to 0.00, 0.58, 0.76, respectively and balanced accuracy score corresponding to 0.44), with scores decreasing for models incorporating windows that included additional races further from the event. Feature importance analysis of ML models demonstrated that retirement was predicted by older age, poor performance, and longer racing career, enforced rest was predicted by younger age and better performance, but was less likely to occur when the stride length is increasing, and MSI predicted by increased number of starters, greater variation in speed and lower percentage of career time at rest. A relatively low classification performance highlights the difficulties in discerning MSI from alternate events using ML. Improved data recording through more thorough assessment and annotation of adverse events is expected to improve the predictability of MSI.
© 2024. The Author(s).
Get Full Text
Publication Date: 2024-11-22 PubMed ID: 39578597PubMed Central: PMC11584853DOI: 10.1038/s41598-024-79071-1Google Scholar: Lookup
The Equine Research Bank provides access to a large database of publicly available scientific literature. Inclusion in the Research Bank does not imply endorsement of study methods or findings by Mad Barn.
LinkedInCopy
  • Journal Article

Summary

This research summary has been generated with artificial intelligence and may contain errors and omissions. Refer to the original study to confirm details provided. Submit correction.

The research used a machine learning model to predict musculoskeletal injuries, enforced rest, and retirement events in racehorses based on stride characteristics. The results showed that certain factors like speed, stride length, and racing history could predict these events to some extent, but the overall prediction performance was still low due to limitations in data recording.

Approach and Methodology

  • The researchers used a machine learning approach to study stride characteristics, and their change over time, in Thoroughbred racehorses.
  • Specifically, a CatBoost model was used to predict the occurrence of musculoskeletal injuries (MSI), enforced rest, and retirement events.
  • The model used features from two races prior to an event and aimed to predict these outcomes based on stride characteristics at the horseracing level.

Results

  • The model’s highest classification performance was obtained when using data from the two races just before an event (sensitivity score for MSI, enforced rest and retirement equal to 0.00, 0.58, 0.76, respectively).
  • In terms of feature importance, the likelihood of retirement increased with older age, poorer performance, and longer racing careers. Enforced rest was more likely in younger horses with better performance, but less likely when the stride length increased. Increased number of starters, greater variation in speed, and lower percentage of career time at rest were predictors for MSI.

Implications

  • The research highlights the potential of using machine learning to predict injuries and performance in racehorses, but also underscores the challenges in collecting and interpreting the necessary data.
  • The overall classification performance of the model was relatively low, suggesting that it is difficult to accurately predict MSI or other events in racehorses using only stride characteristics data.
  • The researchers suggested that more thorough data recording and the use of more comprehensive feature sets may help to improve the predictability of adverse events in racehorses.

Cite This Article

APA
Bogossian PM, Nattala U, Wong ASM, Morrice-West AV, Zhang GZ, Rana P, Whitton RC, Hitchens PL. (2024). A machine learning approach to identify stride characteristics predictive of musculoskeletal injury, enforced rest and retirement in Thoroughbred racehorses. Sci Rep, 14(1), 28967. https://doi.org/10.1038/s41598-024-79071-1

Publication

Scientific reports
ISSN: 2045-2322
NlmUniqueID: 101563288
Country: England
Language: English
Volume: 14
Issue: 1
Pages: 28967
PII: 28967

Researcher Affiliations

Bogossian, Paulo M
  • Veterinary School, City University of Sao Caetano Do Sul, 30 Santo Antonio St Sao Caetano Do Sul SP, São Caetano do Sul, Brazil.
  • Equine Centre, Melbourne Veterinary School, The University of Melbourne, 250 Princes Hwy Werribee, Melbourne, VIC, 3030, Australia.
Nattala, Usha
  • Melbourne Data Analytics Platform, The University of Melbourne, 700 Swanston Street, Carlton, VIC, 3053, Australia.
Wong, Adelene S M
  • Equine Centre, Melbourne Veterinary School, The University of Melbourne, 250 Princes Hwy Werribee, Melbourne, VIC, 3030, Australia.
Morrice-West, Ashleigh V
  • Equine Centre, Melbourne Veterinary School, The University of Melbourne, 250 Princes Hwy Werribee, Melbourne, VIC, 3030, Australia.
Zhang, Geordie Z
  • Melbourne Data Analytics Platform, The University of Melbourne, 700 Swanston Street, Carlton, VIC, 3053, Australia.
Rana, Pratibha
  • Melbourne Data Analytics Platform, The University of Melbourne, 700 Swanston Street, Carlton, VIC, 3053, Australia.
Whitton, R Chris
  • Equine Centre, Melbourne Veterinary School, The University of Melbourne, 250 Princes Hwy Werribee, Melbourne, VIC, 3030, Australia.
Hitchens, Peta L
  • Equine Centre, Melbourne Veterinary School, The University of Melbourne, 250 Princes Hwy Werribee, Melbourne, VIC, 3030, Australia. peta.hitchens@unimelb.edu.au.

MeSH Terms

  • Horses
  • Machine Learning
  • Animals
  • Rest / physiology
  • Running
  • Musculoskeletal Diseases / veterinary
  • Musculoskeletal System / injuries
  • Horse Diseases / diagnosis
  • Male

Conflict of Interest Statement

Declarations. Competing interests: The authors declare no competing interests. Ethical animal research: Data were sourced from existing collections of data and involved no direct work with animals. The Animal Ethics Committee at the University of Melbourne Faculty of Veterinary and Agricultural Science gave an exemption for formal ethics approval.

References

This article includes 50 references
  1. Johnson BJ. Causes of death in racehorses over a 2 year period. Equine Vet. J. 26, 327–330 (1994).
    pubmed: 8575402doi: 10.1111/j.2042-3306.1994.tb04395.xgoogle scholar: lookup
  2. Flash ML, Renwick M, Gilkerson JR, Stevenson MA. Descriptive analysis of Thoroughbred horses born in Victoria, Australia, in 2010; barriers to entering training and outcomes on exiting training and racing. PLoS One 15(10), e0241273 (2020).
    pmc: PMC7592779pubmed: 33112903doi: 10.1371/journal.pone.0241273google scholar: lookup
  3. Shrestha K, Gilkerson JR, Stevenson MA, Flash ML. Drivers of exit and outcomes for Thoroughbred racehorses participating in the 2017–2018 Australian racing season. PLoS One 16(9), e0257581 (2021).
    pmc: PMC8454983pubmed: 34547036doi: 10.1371/journal.pone.0257581google scholar: lookup
  4. Thomson PC, Hayek AR, Jones B, Evans DL, McGreevy PD. Number, causes and destinations of horses leaving the Australian Thoroughbred and Standardbred racing industries. Aust. Vet. J. 92, 303–311 (2014).
    pubmed: 24954530doi: 10.1111/avj.12204google scholar: lookup
  5. Heleski C. Thoroughbred Racehorse Welfare through the Lens of ‘Social License to Operate—With an Emphasis on a U.S. Perspective. Sustainability 12, 1706 (2020).
    doi: 10.3390/sህ1706google scholar: lookup
  6. Hitchens PL, Morrice-West AV, Stevenson MA, Whitton RC. Meta-analysis of risk factors for racehorse catastrophic musculoskeletal injury in flat racing. Vet. J. 245, 29–40 (2019).
    pubmed: 30819423doi: 10.1016/j.tvjl.2018.11.014google scholar: lookup
  7. Boden LA. Sudden death in racing Thoroughbreds in Victoria Australia. Equine Vet. J. 37, 269–271 (2005).
    pubmed: 15892239doi: 10.2746/0425164054530597google scholar: lookup
  8. Colgate VA, Marr CM. Science-in-brief: Risk assessment for reducing injuries of the fetlock bones in Thoroughbred racehorses. Equine Vet. J. 52, 482–488 (2020).
    pubmed: 32525619doi: 10.1111/evj.13273google scholar: lookup
  9. Navas de Solis C, Gabbett T, King MR, Keene R, McKenzie E. Science in brief: The Dorothy Havemeyer International Workshop on poor performance in horses: Recent advances in technology to improve monitoring and quantification. Equine Vet. J. 54, 844–846 (2022).
    pubmed: 35905088doi: 10.1111/evj.13608google scholar: lookup
  10. Tranquille CA, Murray RC, Parkin TD. Can we use subchondral bone thickness on high-field magnetic resonance images to identify Thoroughbred racehorses at risk of catastrophic lateral condylar fracture?. Equine Vet. J. 49, 167–171 (2017).
    pubmed: 27030308doi: 10.1111/evj.12574google scholar: lookup
  11. Jackson BF. Bone biomarkers and risk of fracture in two- and three-year-old Thoroughbreds. Equine Vet. J. 41, 410–413 (2009).
    pubmed: 19562906doi: 10.2746/042516409x416206google scholar: lookup
  12. Blott SC. A genome-wide association study demonstrates significant genetic variation for fracture risk in Thoroughbred racehorses. BMC Genom. 15, 147 (2014).
    pmc: PMC4008154pubmed: 24559379doi: 10.1186/1471-2164-15-147google scholar: lookup
  13. Darbandi H, Munsters C, Parmentier J, Havinga P. Detecting fatigue of sport horses with biomechanical gait features using inertial sensors. PLoS One 18(4), e0284554 (2023).
    pmc: PMC10104328pubmed: 37058516doi: 10.1371/journal.pone.0284554google scholar: lookup
  14. Peham C, Licka T, Girtler D, Scheidl M. The influence of lameness on equine stride length consistency. Vet. J. 162, 153–157 (2001).
    pubmed: 11531399doi: 10.1053/tvjl.2001.0593google scholar: lookup
  15. Wong ASM, Morrice-West AV, Whitton RC, Hitchens PL. Changes in Thoroughbred speed and stride characteristics over successive race starts and their association with musculoskeletal injury. Equine Vet. J. 55, 194–204 (2023).
    pmc: PMC10084173pubmed: 35477925doi: 10.1111/evj.13581google scholar: lookup
  16. Morrice-West AV. Variation in GPS and accelerometer recorded velocity and stride parameters of galloping Thoroughbred horses. Equine Vet. J. 53, 1063–1074 (2021).
    pubmed: 33098592doi: 10.1111/evj.13370google scholar: lookup
  17. Riggs CM, Whitehouse GH, Boyde A. Pathology of the distal condyles of the third metacarpal and third metatarsal bones of the horse. Equine Vet. J. 31, 140–148 (1999).
    pubmed: 10213426doi: 10.1111/j.2042-3306.1999.tb03807.xgoogle scholar: lookup
  18. Stover SM. An association between complete and incomplete stress fractures of the humerus in racehorses. Equine Vet. J. 24, 260–263 (1992).
    pubmed: 1499531doi: 10.1111/j.2042-3306.1992.tb02831.xgoogle scholar: lookup
  19. Whitton RC. Third metacarpal condylar fatigue fractures in equine athletes occur within previously modelled subchondral bone. Bone 47, 826–831 (2010).
    pubmed: 20659599doi: 10.1016/j.bone.2010.07.019google scholar: lookup
  20. Kokkotis C. Leveraging explainable machine learning to identify gait biomechanical parameters associated with anterior cruciate ligament injury. Sci. Rep. 12, 6647 (2022).
    pmc: PMC9026057pubmed: 35459787doi: 10.1038/s41598-022-10666-2google scholar: lookup
  21. Vallmuur K. Harnessing information from injury narratives in the “big data” era: understanding and applying machine learning for injury surveillance. Inj. Prev. 22(Suppl 1), i34-42 (2016).
    pmc: PMC4852152pubmed: 26728004doi: 10.1136/injuryprev-2015-041813google scholar: lookup
  22. Mouloodi S. What can artificial intelligence and machine learning tell us? A review of applications to equine biomechanical research. J Mech. Behav. Biomed. Mater. 123, 104728 (2021).
    pubmed: 34412024doi: 10.1016/j.jmbbm.2021.104728google scholar: lookup
  23. Schobesberger H, Peham C. Computerized detection of supporting forelimb lameness in the horse using an Artificial Neural Network. Vet. J. 163, 77–84 (2002).
    pubmed: 11749140doi: 10.1053/tvjl.2001.0608google scholar: lookup
  24. Keegan KG, Arafat S, Skubic M, Wilson DA, Kramer J. Detection of lameness and determination of the affected forelimb in horses by use of continuous wavelet transformation and neural network classification of kinematic data. Am. J. Vet. Res. 64, 1376–1381 (2003).
    pubmed: 14620773doi: 10.2460/ajvr.2003.64.1376google scholar: lookup
  25. Eerdekens A. Automatic equine activity detection by convolutional neural networks using accelerometer data. Comput. Electron. Agricult. 168, 105139 (2020).
    doi: 10.1016/j.compag.2019.105139google scholar: lookup
  26. Mouloodi S, Rahmanpanah H, Burvill C, Gohari S, Davies HMS. Experimental, regression learner, numerical, and artificial neural network analyses on a complex composite structure subjected to compression loading. Mech. Adv. Mater. Struct. 29, 2437–2453 (2022).
    doi: 10.1080/15376494.2020.1864071google scholar: lookup
  27. Rahmanpanah H, Mouloodi S, Burvill C, Gohari S, Davies HMS. Prediction of load-displacement curve in a complex structure using artificial neural networks: A study on a long bone. Int. J. Eng. Sci. 54, 103319 (2020).
    doi: 10.1016/j.ijengsci.2020.103319google scholar: lookup
  28. Morrice-West AV, Hitchens PL, Walmsley EA, Stevenson MA, Whitton RC. Training practices, speed and distances undertaken by Thoroughbred racehorses in Victoria Australia. Equine Vet. J. 52, 273–280 (2020).
    pubmed: 31386764doi: 10.1111/evj.13156google scholar: lookup
  29. Crowther MJ, Lambert PC. Stgenreg: A Stata Package for General Parametric Survival Analysis. J. Statis. Soft. 53, 1–17 (2013).
    doi: 10.18637/jss.v053.i12google scholar: lookup
  30. Ke G. LightGBM: a highly efficient gradient boosting decision tree. In Proceedings of the 31st International Conference on Neural Information Processing Systems. 3149–3157 (2017).
  31. Bentéjac C, Csörgő A, Martínez-Muñoz G. A comparative analysis of gradient boosting algorithms. Artif. Intell. Rev. 54, 1937–1967 (2021).
    doi: 10.1007/s10462-020-09896-5google scholar: lookup
  32. Belle V, Papantonis I. Principles and practice of explainable machine learning. Front Big Data 4, 688969 (2021).
    pmc: PMC8281957pubmed: 34278297doi: 10.3389/fdata.2021.688969google scholar: lookup
  33. Parsa AB, Movahedi A, Taghipour H, Derrible S, Mohammadian A. Toward safer highways, application of XGBoost and SHAP for real-time accident detection and feature analysis. Accident Anal. Prevent. (2020).
    pubmed: 31864931doi: 10.1016/j.aap.2019.105405google scholar: lookup
  34. Grandini M, Bagli E, Visani G. Metrics for Multi-Class Classification: an Overview. ArXiv 1–17 (2020).
  35. Nattala U. Verification.Tools: A web tool to evaluate the accuracy of predictions/forecasts. (2023).
  36. Bailey CJ, Rose RJ, Reid SW, Hodgson DR. Wastage in the Australian Thoroughbred racing industry: a survey of Sydney trainers. Aust. Vet. J. 75, 64–66 (1997).
    pubmed: 9034505doi: 10.1111/j.1751-0813.1997.tb13836.xgoogle scholar: lookup
  37. Kim B, Kim J. Adjusting Decision Boundary for Class Imbalanced Learning. IEEE Access 8, 81674–81685 (2020).
    doi: 10.1109/access.2020.2991231google scholar: lookup
  38. Hill AE, Gardner IA, Carpenter TE, Stover SM. Effects of injury to the suspensory apparatus, exercise, and horseshoe characteristics on the risk of lateral condylar fracture and suspensory apparatus failure in forelimbs of Thoroughbred racehorses. Am. J. Vet. Res. 65, 1508–1517 (2004).
    pubmed: 15566089doi: 10.2460/ajvr.2004.65.1508google scholar: lookup
  39. Anthenill LA, Stover SM, Gardner IA, Hill AE. Risk factors for proximal sesamoid bone fractures associated with exercise history and horseshoe characteristics in Thoroughbred racehorses. Am. J. Vet. Res. 68, 760–771 (2007).
    pubmed: 17605612doi: 10.2460/ajvr.68.7.760google scholar: lookup
  40. Vallance SA, Entwistle RC, Hitchens PL, Gardner IA, Stover SM. Case–control study of high-speed exercise history of Thoroughbred and Quarter Horse racehorses that died related to a complete scapular fracture. Equine Vet. J. 45, 284–292 (2013).
    pubmed: 23240828doi: 10.1111/j.2042-3306.2012.00644.xgoogle scholar: lookup
  41. Martig S, Chen W, Lee PVS, Whitton RC. Bone fatigue and its implications for injuries in racehorses. Equine Vet. J. 46, 408–415 (2014).
    pubmed: 24528139doi: 10.1111/evj.12241google scholar: lookup
  42. Porr CA, Kronfeld DS, Lawrence LA, Pleasant RS, Harris PA. Deconditioning reduces mineral content of the third metacarpal bone in horses. J. Anim. Sci. 76, 1875–1879 (1998).
    pubmed: 9690643doi: 10.2527/1998.7671875xgoogle scholar: lookup
  43. Carrier TK. Association between long periods without high-speed workouts and risk of complete humeral or pelvic fracture in Thoroughbred racehorses: 54 cases (1991–1994). J. Am. Vet. Med. Assoc. 212, 1582–1587 (1998).
    pubmed: 9604029doi: 10.2460/javma.1998.212.10.1582google scholar: lookup
  44. Parkin TD. Race- and course-level risk factors for fatal distal limb fracture in racing Thoroughbreds. Equine Vet. J. 36, 521–526 (2004).
    pubmed: 15460077doi: 10.2746/0425164044877332google scholar: lookup
  45. Parkin TD. Risk factors for fatal lateral condylar fracture of the third metacarpus/metatarsus in UK racing. Equine Vet. J. 37, 192–199 (2005).
    pubmed: 15892225doi: 10.2746/0425164054530641google scholar: lookup
  46. Dong J. Machine learning model for early prediction of acute kidney injury (AKI) in pediatric critical care. Crit. Care 25, 288 (2021).
    pmc: PMC8353807pubmed: 34376222doi: 10.1186/s13054-021-03724-0google scholar: lookup
  47. Prokhorenkova L, Gusev G, Vorobev A, Veronika Dorogush A, Gulin A. CatBoost: unbiased boosting with categorical features. Advances in Neural Information Processing Systems 31, 466 (2018).
  48. Parkes RSV, Weller R, Pfau T, Witte TH. The Effect of Training on Stride Duration in a Cohort of Two-Year-Old and Three-Year-Old Thoroughbred Racehorses. Animals (Basel) 9 (2019).
    pmc: PMC6680649pubmed: 31336595doi: 10.3390/ani9070466google scholar: lookup
  49. Schrurs C, Blott S, Dubois G, Van Erck-Westergren E, Gardner DS. Locomotory profiles in Thoroughbreds: Peak stride length and frequency in training and association with race outcomes. Animals (Basel) 12 (2022).
    pmc: PMC9741461pubmed: 36496790doi: 10.3390/ani12233269google scholar: lookup
  50. Takahashi Y, Takahashi T, Mukai K, Ohmura H. Effects of fatigue on stride parameters in Thoroughbred racehorses during races. J. Equine Vet. Sci. 101, 103447 (2021).
    pubmed: 33993952doi: 10.1016/j.jevs.2021.103447google scholar: lookup

Citations

This article has been cited 0 times.
Subscribe to our newsletter
Join 99,344 horse owners like you who receive our email updates on horse health and nutrition.