Horses are amazing athletes, predisposed to musculoskeletal injury (MSI) due to high workloads incurred during training and competition. The development of wearable technology and optical motion capture (OMC) systems has enabled the collection of large volumes of high-resolution data on the workloads and biomechanics of equine athletes. This volume of data, however, has generated challenges in timely data analyses and interpretation, providing scope for harnessing artificial intelligence (AI) in MSI prevention. We aimed to systematically describe the application of AI models to wearable sensor data and motion capture systems for gait analysis, and their association with MSI in equine athletes. Due to limited equine literature, human athletes were also included. Following the PRISMA guidelines, a total of 1217 peer-reviewed journal articles and conference proceeding were identified between the years 2000 to 2024, and 33 studies (19 equine, 14 human) met the eligibility criteria and were included in this review.
Two methodologies for approaching injury prevention with AI were identified: predictive MSI models and classifying gait patterns in the context of lameness. For the studies classifying gait patterns, two subdivisions in equine athletes were recognised: established (quantifying asymmetry at the trot) and unestablished methods (lameness at the walk or gallop, hoof deformation). Three predictive MSI models in human athletes were identified, all prospective studies with only one using longitudinal training and competition data from wearable devices. The other two used cross-sectional wearable and/or OMC data from protocol-based fitness tests. In equines, a single publication using AI on retrospective longitudinal race-day wearable data was identified, but whilst having some success in predicting enforced rest and retirement, it was poor at predicting MSI. Predictive models are a promising screening tool for injury prevention but are clearly at a developmental stage. Further research is critical to assess their reliability and capacity for injury prediction.