Fig 1.
Summary of health condition distribution and medical history characteristics.
(a) Boxplot comparing each dog’s age at last reported health condition with the total span of their medical history (i.e., time between first and last reported condition). (b) Histogram representing number of dogs with different numbers of conditions. (c) Bar chart displaying the 20 most frequently occurring health conditions in the study population.
Fig 2.
Associations between demographic factors and health conditions, stratified by condition category.
Coefficients from logistic regression models are shown for each health condition, with age, weight, breed background, and sex/reproductive status as predictors. Statistical significance is denoted by asterisks (* p < 0.05, ** p < 0.01, *** p < 0.001).
Fig 3.
Undirected comorbidity network.
Nodes are health conditions with node size proportional to condition prevalence, and edges are statistically significant comorbidity connections (at a Bonferroni adjusted P-value < 0.001). Colors indicate health condition categories. (a) Full network view with Bone/Orthopedic, Ear/Nose/Throat, Eye, Gastrointestinal, Infection/Parasites, Mouth/Dental/Oral, Skin, and Respiratory categories collapsed into single aggregate nodes. (b) Zoomed-in view of the Skin conditions, including their first-degree neighbors to ear infection.
Fig 4.
Age-stratified comorbidity networks.
Undirected comorbidity networks in Young adult (a), Mature adult (b) and Senior (c). In each network, nodes represent health conditions with node size proportional to condition prevalence, and edges represent statistically significant comorbidity associations (at a Bonferroni-adjusted P-value < 0.01). Colors indicate health condition categories. No network is shown for the Puppy stratum as no significant comorbidities were detected in this age group. Conditions in the Skin and Infection/Parasites categories are presented as aggregate nodes in (b) and (c) to minimize cluttering. (d) Heatmap displaying edge overlap between disease networks across age groups. Numbers indicate shared edges between a pair of networks, with diagonal values showing total number of edges per network.
Table 1.
Network topology measures across life stages: edge density refers to the number of edges relative to the maximum number of possible edges in each graph; clustering coefficient measures the degree to which nodes in each graph cluster together; betweenness and degree centrality refer to the normalized graph level centralization measure based on the respective node-level centrality scores.
Fig 5.
Time-directed network using a window size of 12 months.
Nodes are health conditions, and edges are statistically significant comorbidity connections (at a Bonferroni-adjusted P-value < 0.01). Colors indicate health condition categories. Arrowheads point from the health condition that occurs earlier in time to the health condition that occurs later. (a) Full network view with Skin and Infection/Parasites categories collapsed into single aggregate nodes. (b) Zoomed-in view of the Skin and Infection/Parasites conditions, including their first-degree neighbors.