A chronological age prediction model using clinical biochemistry and hematological data identifies features of biological age in healthy Labrador retrievers
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
We hypothesised that clinical blood biochemistry and haematological data could predict the age of healthy dogs. Using a large veterinary electronic medical record (EMR) database, we investigated age-related relationships in Labrador Retrievers, a common large breed. Analysing over 3 million blood test values across 24 analytes and ~ 145,000 dogs, we modelled age and sex using a generalised additive model. Changes in multiple analytes were seen around 3-6 and 12 years of age. As these findings demonstrated age-related changes in blood analytes, we considered the data suitable for modelling chronological age. Using eXtreme Gradient Boosting, we predicted age based on sex, 24 blood analytes, and 276 derived ratios. Training, test and validation sets for the model consisted of ~ 38.5 k, ~ 4.7 k and ~ 4.7 k data points, respectively. The model achieved a mean absolute error of approximately 14.6 months. We hypothesised that some aspect of differences between predicted and chronological age might represent features of biological age, linked to the onset of age-related diseases. We demonstrated that > 6-month difference in predicted and chronological age was significantly associated with the first diagnosis of common age-related conditions such as arthritis, lumps/growths, nuclear sclerosis and hepatopathy. These findings suggest that standard clinical measures from a large EMR database can be used to predict age in healthy pets and also provide insights into biological age, aiding in the assessment of individual health risks.
Fields of Study & Subjects
Veterinary Medicine and Surgery
Genetics
Aging
and Longevity in Model Organisms
Human-Animal Interaction Studies
Publication details
- Journal / source
- GeroScience
- Publication date
- 28 March 2026
- Language
- EN