The Effect of Training on Stride Duration in a Cohort of Two-Year-Old and Three-Year-Old Thoroughbred Racehorses

Field-training in young two-year-old thoroughbreds: investigating cardiorespiratory adaptations and the presence of exercise induced pulmonary hemorrhage

BACKGROUND

Comparatively little is known regarding the initial cardiorespiratory response of young racehorses to training. The objectives were to compare physiological parameters before and after introductory training and determine whether young Thoroughbreds show endoscopic signs of exercise-induced pulmonary hemorrhage (EIPH). Ten Thoroughbreds (20-23 months) underwent 12-weeks of introductory training, including weekly speed sessions. Two 600 m high-speed exercise tests (HSET) were performed following weeks 4 and 12 while wearing a validated ergospirometry facemask. Peak oxygen consumption (V̇O2pk) and ventilatory parameters (tidal volume, VT; peak inspiratory and expiratory flow, PkV̇I, PkV̇E; respiratory frequency, Rf; minute ventilation, V̇E) were measured. The ventilatory equivalent of oxygen (V̇E/V̇O2) and the aerobic and anaerobic contributions to energy production were calculated. Maximal heart rate (HRmax) and HR at maximal speed (HRVmax) were determined. Post-exercise hematocrit, plasma ammonia and blood lactate were measured. Evidence of EIPH was investigated via tracheobronchoscopy post-exercise. Results were compared (paired t-test, P < 0.05).

RESULTS

Horses were faster following training (P < 0.001) and V̇O2pk increased 28 ml/(kg total mass.min) (28 ± 16%; P < 0.001). Ventilatory (V̇E, P = 0.0015; Rf, P < 0.001; PkV̇I, P < 0.001; PkV̇E, P < 0.001) and cardiovascular parameters (HRmax, P = 0.03; HRVmax, P = 0.04) increased. The increase in V̇E was due to greater Rf, but not VT. V̇E/V̇O2 was lower (26 ± 3.6 vs 23 ± 3.7; P = 0.02), indicating improved ventilatory efficiency. Anaerobic contribution to total energy production increased from 15.6 ± 6.1% to 18.5 ± 6.3% (P = 0.02). Post-exercise hematocrit (P < 0.001), plasma ammonia (P = 0.03) and blood lactate (P = 0.001) increased following training. Horses showed no signs of EIPH.

CONCLUSIONS

Young two-year-old Thoroughbreds responded well to introductory training without developing tracheobronchoscopic evidence of EIPH.

Dirt Track Surface Preparation and Associated Differences in Speed, Stride Length, and Stride Frequency in Galloping Horses

In racehorses, the risk of musculoskeletal injury is linked to a decrease in speed and stride length (SL) over consecutive races prior to injury. Surface characteristics influence stride parameters. We hypothesized that large changes in stride parameters are found during galloping in response to dirt racetrack preparation. Harrowing of the back stretch of a half-mile dirt racetrack was altered in three individual lanes with decreasing depth from the inside to the outside. Track underlay compaction and water content were changed between days. Twelve horses (six on day 2) were sequentially galloped at a target speed of 16 ms-1 across the three lanes. Speed, stride frequency (SF), and SL were quantified with a GPS/GNSS logger. Mixed linear models with speed as covariate analyzed SF and SL, with track hardness and moisture content as fixed factors (p < 0.05). At the average speed of 16.48 ms-1, hardness (both p < 0.001) and moisture content (both p < 0.001) had significant effects on SF and SL. The largest difference in SL of 0.186 m between hardness and moisture conditions exceeded the 0.10 m longitudinal decrease over consecutive race starts previously identified as injury predictor. This suggests that detailed measurements of track conditions might be useful for refining injury prediction models.

Associations between Racing Thoroughbred Movement Asymmetries and Racing and Training Direction

BACKGROUND

Racehorses commonly train and race in one direction, which may result in gait asymmetries. This study quantified gait symmetry in two cohorts of Thoroughbreds differing in their predominant exercising direction; we hypothesized that there would be significant differences in the direction of asymmetry between cohorts.

METHODS

307 Thoroughbreds (156 from Singapore Turf Club (STC)-anticlockwise; 151 from Hong Kong Jockey Club (HKJC)-clockwise) were assessed during a straight-line, in-hand trot on firm ground with inertial sensors on their head and pelvis quantifying differences between the minima, maxima, upward movement amplitudes (MinDiff, MaxDiff, UpDiff), and hip hike (HHD). The presence of asymmetry (≥5 mm) was assessed for each variable. Chi-Squared tests identified differences in the number of horses with left/right-sided movement asymmetry between cohorts and mixed model analyses evaluated differences in the movement symmetry values.

RESULTS

HKJC had significantly more left forelimb asymmetrical horses (Head: MinDiff p < 0.0001, MaxDiff p < 0.03, UpDiff p < 0.01) than STC. Pelvis MinDiff (p = 0.010) and UpDiff (p = 0.021), and head MinDiff (p = 0.006) and UpDiff (p = 0.017) values were significantly different between cohorts; HKJC mean values indicated left fore- and hindlimb asymmetry, and STC mean values indicated right fore- and hindlimb asymmetry.

CONCLUSION

the asymmetry differences between cohorts suggest that horses may adapt their gait to their racing direction, with kinematics reflecting reduced 'outside' fore- and hindlimb loading.

Circle Diameter Impacts Stride Frequency and Forelimb Stance Duration at Various Gaits in Horses

The effects of gait and diameter have been studied independently, but rarely together in equine circular exercise studies. This study aimed to determine the impact of diameter (10-m or 15-m) at various gaits (walk, trot, and canter) on stride frequency or forelimb stance duration. Nine mature horses were outfitted with Tekscan™ Hoof Sensors on their forelimbs during circular and straight-line exercise at various gaits on a clay and sand arena surface. Statistical analysis was performed in SAS 9.4 with fixed effects of exercise type, recording, leg, and breed (PROC GLIMMIX, p < 0.05 significance). At walk (p < 0.0001) and trot (p < 0.001), stride frequency was lower during circular exercise. Stride frequency was similar between forelimbs at all gaits. At walk (p < 0.001) and canter (p = 0.01), stance duration was greatest during 10-m circle exercise. At walk (p = 0.0007), trot (p < 0.001), and canter (p < 0.0001), the inside forelimb had longer stance duration than the outside forelimb. Differences between forelimb stance durations may support asymmetrical travel while horses exercise on a circle at the walk, trot, and canter. These results demonstrate diameter and gait are important factors when evaluating forelimb kinematics during circular exercise.

Rein tension in harness trotters during on-track exercise

Horseracing is under public scrutiny with increasing demands to safeguard horse welfare. It is accepted that, as a result of bit pressure and/or equipment, mouth lesions accompany many types of horse use, including racing. However, there are currently no data available on the range of bit pressures in driven trotters. Our aim was to investigate whether rein tension (RT, proxy for bit pressures) differs among gaits, between tempo within gait, between horses and drivers, and between left/right reins. Standardbreds (n = 9), driven by experienced drivers (n = 11), performed exercise tests on a racetrack (cross-over design; total 31 tests, data available from 26 tests). Horses' motion symmetry was measured before tests (trotting in hand). Rein tension, speed and heart rate were measured during exercise. A moving-window filter was applied to RT raw data. Median, maximum and interquartile range for the estimated stride median RT were determined for each rein (left/right) and segment: walk; circling in slow trot followed by transition to faster trot; fast (racing) trot; and slowing down to walk. Mixed models were used for statistical analysis. Least square means for segment median RT ranged between 17-19 N in walk, 34-40 N during circling-accelerating, 51-62 N in fast trot, and 53-71 N for slowing down. Segment maximum RT was between 60-81 N in walk, 104-106 N during circling-accelerating, 72-86 N in fast trot, and 86-129 N during slowing down. Interquartile ranges were between 7-9 N in walk, 28-31 N during circling-accelerating, 8-10 N in fast trot, and 12-18 N for slowing down. Hind limb asymmetry exceeded the recommended threshold in three horses and was associated with higher median (48 N) and maximum (106 N) RT than symmetric horses (29 N and 73 N, respectively, p < 0.01). Consistent left-right asymmetry in RT was more common among horses than among drivers. Rein tension increased with increasing heart rate (p ≤ 0.0006). Rein tensions were higher than those reported during riding or in horses worked from the ground. The findings of high RT, taken together with the high reported prevalence of oral injuries in harness trotters, call for further research into RT, motion symmetry and use of equipment.

Quantitative gait analysis before and after a cross-country test in a population of elite eventing horses

Early recognition of lameness is crucial for injury prevention. Quantitative gait analysis systems can detect low-grade asymmetries better than the human eye and may be useful in early lameness recognition. The aims of this study were 1) to investigate the frequency of gait asymmetries based on head and pelvic movement in elite eventing horses using inertial mounted measurement units and 2) to assess the association between asymmetries and muscle enzymes and blood lactate (LA) levels post exercise. Movement asymmetry of the head, wither, and pelvis were quantified in 33 elite eventing horses prior to and one day after the cross-country test of three Concours Complet International (CCI3* and CCI4*) events held three weeks apart. The effects of LA concentration immediately after completion of the cross-country course and of serum levels of creatine kinase (CK) and aspartate amino-transferase (AST) four hours post-exercise on gait asymmetry parameters were tested with linear models. A total of 58% and 77% of the 33 horses exhibited gait asymmetries that exceeded published threshold values before and after the cross-country course, respectively. The magnitude of pre-existing gait asymmetries was not significantly increased after the cross-country test and no associations with post-exercise levels of CK, AST, or LA were detected. The stride duration was significantly shorter the day following the cross-country test and was associated with LA, the age and the weight of the horses. In conclusion, a majority of the horses studied presented gait asymmetries and strenuous exercise resulted in decreased stride duration but did not worsen gait asymmetries.

Relationship between Thoroughbred workloads in racing and the fatigue life of equine subchondral bone

Fatigue life (FL) is the number of cycles of load sustained by a material before failure, and is dependent on the load magnitude. For athletes, ‘cycles’ translates to number of strides, with load proportional to speed. To improve previous investigations estimating workload from distance, we used speed (m/s, x) per stride collected using 5 Hz GPS/800 Hz accelerometer sensors as a proxy for limb load to investigate factors associated with FL in a Thoroughbred race start model over 25,234 race starts, using a combination of mathematical and regression modelling. Fore-limb vertical force (NKg^-1) was estimated using a published equation: Vertical force = 2.778 + 2.1376x − 0.0535x². Joint load (σ) was estimated based on the vertical force, scaled according to the maximum speed and defined experimental loads for the expected variation in load distribution across a joint surface (54-90 MPa). Percentage FL (%FL) was estimated using a published equation for cycles to failure (N_f) summed across each race start: N_f = 10^{(σ-134.2)/−14.1.} Multivariable mixed-effects linear regression models were generated on %FL, adjusting for horse-level clustering, presented as coefficients; 95%CI. Scaled to the highest joint load, individual starts accrued a mean of 9.34%FL (sd. 1.64). Older age (coef. 0.03; 0.002–0.04), longer race-distances (non-linear power transformed), and firmer track surfaces (ref. Heavy 10: Good 3 coef. 2.37; 2.26–2.48) were associated with greater %FL, and males accrued less than females (p < 0.01). Most variables associated with %FL are reported risk factors for injury. Monitoring strides in racehorses may therefore allow identification of horses at risk, enabling early detection of injury.

Impact of Gait and Diameter during Circular Exercise on Front Hoof Area, Vertical Force, and Pressure in Mature Horses

Circular exercise can be used at varying gaits and diameters to exercise horses, with repeated use anecdotally relating to increased lameness. This work sought to characterize mean area, mean vertical force, and mean pressure of the front hooves while exercising in a straight line at the walk and trot, and small (10-m diameter) and large circles (15-m diameter) at the walk, trot, and canter. Nine mature horses wore TekscanTM Hoof Sensors on their forelimbs adhered with a glue-on shoe. Statistical analysis was performed in SAS 9.4 with fixed effects of leg, gait, and exercise type (PROC GLIMMIX) and p < 0.05 as significant. For all exercise types, the walk had greater mean pressure than the trot (p < 0.01). At the walk, the straight line had greater mean area loaded than the large circle (p = 0.01), and both circle sizes had lower mean vertical force than the straight line (p = 0.003). During circular exercise at the canter, the outside front limb had greater mean area loaded than at the walk and trot (p = 0.001). This study found that gait is an important factor when evaluating circular exercise and should be considered when exercising horses to prevent injury.

Cross-Modality Interaction Network for Equine Activity Recognition Using Imbalanced Multi-Modal Data

With the recent advances in deep learning, wearable sensors have increasingly been used in automated animal activity recognition. However, there are two major challenges in improving recognition performance—multi-modal feature fusion and imbalanced data modeling. In this study, to improve classification performance for equine activities while tackling these two challenges, we developed a cross-modality interaction network (CMI-Net) involving a dual convolution neural network architecture and a cross-modality interaction module (CMIM). The CMIM adaptively recalibrated the temporal- and axis-wise features in each modality by leveraging multi-modal information to achieve deep intermodality interaction. A class-balanced (CB) focal loss was adopted to supervise the training of CMI-Net to alleviate the class imbalance problem. Motion data was acquired from six neck-attached inertial measurement units from six horses. The CMI-Net was trained and verified with leave-one-out cross-validation. The results demonstrated that our CMI-Net outperformed the existing algorithms with high precision (79.74%), recall (79.57%), F1-score (79.02%), and accuracy (93.37%). The adoption of CB focal loss improved the performance of CMI-Net, with increases of 2.76%, 4.16%, and 3.92% in precision, recall, and F1-score, respectively. In conclusion, CMI-Net and CB focal loss effectively enhanced the equine activity classification performance using imbalanced multi-modal sensor data.

Using Different Combinations of Body-Mounted IMU Sensors to Estimate Speed of Horses-A Machine Learning Approach

Speed is an essential parameter in biomechanical analysis and general locomotion research. It is possible to estimate the speed using global positioning systems (GPS) or inertial measurement units (IMUs). However, GPS requires a consistent signal connection to satellites, and errors accumulate during IMU signals integration. In an attempt to overcome these issues, we have investigated the possibility of estimating the horse speed by developing machine learning (ML) models using the signals from seven body-mounted IMUs. Since motion patterns extracted from IMU signals are different between breeds and gaits, we trained the models based on data from 40 Icelandic and Franches-Montagnes horses during walk, trot, tölt, pace, and canter. In addition, we studied the estimation accuracy between IMU locations on the body (sacrum, withers, head, and limbs). The models were evaluated per gait and were compared between ML algorithms and IMU location. The model yielded the highest estimation accuracy of speed (RMSE = 0.25 m/s) within equine and most of human speed estimation literature. In conclusion, highly accurate horse speed estimation models, independent of IMU(s) location on-body and gait, were developed using ML.

The effect of curve running on distal limb kinematics in the Thoroughbred racehorse

During racing, injury is more likely to occur on a bend than on a straight segment of track. This study aimed to quantify the effects of galloping at training speeds on large radius curves on stride parameters and limb lean angle in order to assess estimated consequences for limb loading. Seven Thoroughbred horses were equipped with a sacrum-mounted inertial measurement unit with an integrated GPS, two hoof-mounted accelerometers and retro-reflective markers on the forelimbs. Horses galloped 2–4 circuits anticlockwise around an oval track and were filmed at 120 frames per second using an array of ten cameras. Speed and curve radius were derived from GPS data and used to estimate the centripetal acceleration necessary to navigate the curve. Stride, stance and swing durations and duty factor (DF) were derived from accelerometer data. Limb markers were tracked and whole limb and third metacarpus (MCIII) angles were calculated. Data were analysed using mixed effects models with a significance level of p < 0.05. For horses galloping on the correct lead, DF was higher for the inside (lead) leg on the straight and on the curve. For horses galloping on the incorrect lead, there was no difference in DF between inside and outside legs on the straight or on the curve. DF decreased by 0.61% of DF with each 1 m s^-2 increase in centripetal acceleration (p < 0.001). Whole limb inclination angle increased by 1.5° per 1 m s^-1 increase in speed (p = 0.002). Limb lean angles increase as predicted, and lead limb function mirrors the functional requirements for curve running. A more comprehensive understanding of the effects of lean and torque on the distal limb is required to understand injury mechanisms.

Efectos de dos protocolos de entrenamiento sobre el lactato sanguíneo en caballos de paso fino

Objetivo: Comparar los efectos de un protocolo de entrenamiento de resistencia (ER) con un protocolo tradicional (ET) sobre la concentración sanguínea de lactato y CK. Materiales y Métodos: Se aplicaron dos protocolos de entrenamiento durante 6 meses divididos en tres etapas. Se realizaron pruebas de esfuerzo antes de comenzar los protocolos de entrenamiento y al finalizar cada una de las etapas. En cada prueba se tomaron muestras de sangre venosa en reposo, durante el ejercicio y en recuperación para medir el lactato (L), y al inicio y al final para medir la creatin kinasa (CK) e inferir las adaptaciones metabólicas y musculares. Se calculó la diferencia de medianas del lactato basal por medio de la U Mann Whitney y se comparó la diferencia de medias del porcentaje de aclaramiento entre los grupos a través de la T de Students. Resultados: Se encontró una diferencia significativa en el porcentaje de depuración de lactato entre ER y ET. También hubo un aumento significativo de los valores de CK intra-grupos, antes y después de las pruebas, pero manteniéndose dentro de los rangos de referencia. Discusión: El ER aumentó la capacidad de metabolizar el lactato pos-ejercicio en potros con entrenamiento de resistencia, aunque no hubo diferencias entre la máxima producción de L entre el grupo ER y ET. El comportamiento de la CK dentro de los rangos de referencia indica que no hubo daño muscular en los potros de ambos grupos.