Abstract
Canine hip dysplasia is a developmental condition affecting the hip joint, leading to pain, lameness, and osteoarthritis. While radiographic diagnosis is the standard, it has limitations. This pilot study investigates the feasibility of using non-invasive motion sensors and computational methods as a supplementary diagnostic tool. Kinematic (accelerometer) and surface electromyography (sEMG) data were analyzed from two groups of Labrador Retriever dogs: a healthy control group and a dysplastic group. The kinematic signals were processed and compared using the Dynamic Time Warping (DTW) algorithm to measure gait dissimilarity relative to a healthy reference dog. The sEMG data were analyzed with autoregressive (AR) models to extract features of muscle activity. The DTW results demonstrated a statistically significant distinction between the groups, with dysplastic dogs showing a higher dissimilarity cost in their gait. Statistical analysis, including Tukey's HSD test, confirmed that movement patterns within each group were consistent but significantly different between the healthy and dysplastic groups. This study suggests that computational analysis of motion sensor data is a promising approach for identifying gait patterns associated with hip dysplasia, offering potential as an objective and non-invasive auxiliary diagnostic tool. Given the limited sample size, findings should be interpreted cautiously and not generalized beyond this cohort; larger, multi-breed studies are needed to confirm effect sizes, estimate variability, and validate thresholds observed here. Nevertheless, the results indicate promising group separability using DTW on accelerometry and AR features from sEMG.