Objective Assessment of Equine Locomotor Symmetry Using an Inertial Sensor System and Artificial Intelligence: A Comparative Study

Simple Summary Lameness is one of the main causes of poor performance in all equestrian disciplines. Traditionally, it is assessed by clinical assessment, using subjective numerical scales. Quantitative lameness assessment has gained popularity over the last years, aiding in the decision-making process. Recently, a marker-less artificial intelligence (AI) motion tracking system has been developed for lameness assessment, requiring less specialized equipment than traditional methods, along with reduced technical expertise and time-consuming procedures. Our aim was to compare this system with an inertial measurement unit system, as well as with clinical examination, to determine the level of agreement and accuracy of both systems, and their relation to visual examination. In our study, clinical examination detected locomotor asymmetries in accordance with both systems. A greater number of limbs were considered asymmetric by the AI motion tracking system. The highest level of agreement was observed for forelimb movement on a straight line and hard surface, and the lowest for pelvic movement on a straight line and soft surface, probably due to the difficulty in assessing hindlimb asymmetry. It would be interesting to measure locomotor asymmetries regularly in training and conditioning horses, as some degree of asymmetry may be clinically relevant.Abstract In horses, quantitative assessment of gait parameters, as with the use of inertial measurement units (IMUs) systems, might help in the decision-making process. However, it requires financial investment, is time-consuming, and lacks accuracy if displaced. An innovative artificial intelligence marker-less motion tracking system (AI-MTS) may overcome these limitations in the field. Our aim was to compare the level of agreement and accuracy between both systems and visual clinical assessment. Twenty horses underwent locomotion analysis by visual assessment, IMUs, and AI-MTS systems, under the following conditions: straight hard (SH), straight soft (SS), left and right circle hard (LCH, RCH), and soft (LCS, RCS). A greater number of horses were considered sound by clinical examination, compared to those identified as symmetric by the two gait analysis systems. More limbs were considered asymmetric by the AI-MTS compared to IMUs, suggesting its greater sensitivity. The greatest agreement between the two systems was found for the difference between two minima in vertical head position in SH, while the lowest for the difference between two minima in vertical pelvis position in SS, reflecting the difficulties in assessing asymmetry of the hindlimbs. It is unknown what degree of asymmetry is clinically relevant, suggesting that more consistent use in training horses may help determine the thresholds for asymmetry. Some degree of asymmetry may be clinically relevant, suggesting its regular use in training horses.