Determining Objective Parameters to Assess Gait Quality in Franches-Montagnes Horses for Ground Coverage and Over-Tracking - Part 2: At Trot.

In gait quality assessments of horses, stride length (SL) is visually associated with spectacular movements of the front limbs, and described as ground coverage, while the movement of the hind limb under the body is supposedly essential to a longer over-tracking distance (OTD). To identify movement patterns with strong associations to SL and OTD, limb and body kinematics of 24 Franches-Montagnes (FM) stallions were measured with 3D optical motion capture (OMC) on a treadmill during an incremental speed test at trot (3.3-6.5 m/s). These measurements were correlated to the scores of ground coverage and over-tracking from six breeding experts. The amount of explained variance of parameters on SL and OTD were estimated using linear mixed-effect models in two models: a full model with all parameters measurable with OMC, and a reduced model with a subset of parameters measurable with inertial measurement units (IMUs). The front limb stance duration (16%) and OTD (7%) measured with OMC, or the OMC parameters front limb stance duration (24%) and suspension duration (14%) measurable with IMUs explained most variance in SL. However, four of six breeding experts were also significantly correlated (r>|0.41|) to front limb protraction angle. OTD variance was explained with OMC parameters suspension duration (10%) and hind limb contralateral pro-retraction angles (9%) or IMU-measurable parameters suspension duration (20%) and maximal pelvis pitch (5%). Four experts' scores for over-tracking were correlated to suspension duration. These results underscore the need for precise definitions of gait quality traits.

Determining Objective Parameters to Assess Gait Quality in Franches-Montagnes Horses for Ground Coverage and Over-Tracking - Part 1: At Walk.

Ground coverage and over-tracking are two gait quality traits describing the forward movement of the front respectively the hind limbs in relation to stride length and over-tracking distance. To investigate the complex interplay of different movement patterns in ground coverage and over-tracking, limb and body kinematics of 24 Franches-Montagnes (FM) stallions were measured with 3D optical motion capture (OMC) on a treadmill during an incremental speed test at the walk (1.4-2.0 m/s). The significance and amount of explained variance of kinematic parameters on stride length and over-tracking distance were estimated using linear mixed-effect models, with speed and horse as random effects. Two separate models were tested: a full model with all parameters measurable by OMC, and a reduced model with a subset of parameters also measurable with inertial measurement units (IMUs). The kinematic parameters were correlated to the subjective scores from six breeding experts to interpret their external validity. The parameter for ground coverage at the walk, explaining most of the variance in stride length, were the maximal forelimb retraction angle (11%) measured with OMC, and the range of pelvis pitch (10%) if measuring with IMUs. The latter was also the most relevant for quantifying over-tracking, explaining 24% to 33% of the variance in the over-tracking distance. The scores from most breeding experts were significantly correlated (r ≥ |0.41|) with the fore- and hind limb protraction angles, which reflect the textual definition of ground coverage and over-tracking. Both gait quality traits can be objectively quantified using either OMC or IMUs.

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.