Discrimination of the Lame Limb in Horses Using a Machine Learning Method (Support Vector Machine) Based on Asymmetry...

Area of Science:

• Equine biomechanics
• Veterinary diagnostics
• Machine learning applications in animal health

Background:

• Lameness detection in horses is crucial for welfare and performance.
• Mild lameness often presents diagnostic challenges in equine veterinary practice.
• Objective lameness assessment requires advanced analytical tools.

Purpose of the Study:

• To develop a predictive system utilizing a support vector machine (SVM) for identifying the affected limb in horses exhibiting lameness.
• To analyze kinematic data from inertial measurement units (IMUs) for lameness classification.
• To assess the efficacy of SVM in detecting subtle lameness during trot.

Main Methods:

• Data acquisition from inertial measurement units (IMUs) placed on the head, withers, and pelvis of 287 horses (256 lame, 31 sound).
• Analysis of kinematic variables including vertical displacement and retraction angles.
• Development and validation of a support vector machine (SVM) model for lameness classification.

Main Results:

• The SVM model achieved an overall accuracy of 86% in lameness detection.
• Highest accuracy was observed for right and left forelimb lameness detection.
• Challenges included lower accuracy for sound horses (54.8%) and some forelimb-hindlimb misclassifications.

Conclusions:

• The SVM system shows promise for objective, non-invasive lameness detection in horses.
• Vertical head and withers displacement are key variables for accurate lameness classification.
• Future work should explore deep learning and sensor reduction for integrated, early locomotor issue detection systems.