Morphology as a predictor of brachiocephalic obstructive airway syndrome

Brachycephalic obstructive airway syndrome (BOAS) is a respiratory disorder commonly observed in dog breeds with short skulls (Brachycephalic breeds). Although craniofa- cial morphology is widely considered a major risk indicator for BOAS, substantial vari- ation in disease severity exists across brachycephalic breeds. This study investigated the relationship between craniofacial conformation and BOAS risk using conformational measurements from 1,296 dogs across multiple breeds. Five craniofacial-related ratios were analyzed using principal component analysis, linear mixed models, and machine learning approaches. Principal Component Analysis (PCA) revealed strong clustering by breed, suggesting substantial breed-level morphological structure. Linear mixed models indicated that individual conformational traits accounted for only a small proportion of BOAS variation after controlling for breed identity. An XGBoost classifier achieved mod- erate predictive performance (AUC = 0.85). However, the model suggests that many of the predictive signals reflect breed-level differences rather than within-breed variation in craniofacial ratios.