Exploration of the use of sensor data for diagnosis and monitoring of osteoarthritis in companion animals

Abstract

Canine osteoarthrits is a chronic disease, diagnosed in 80% of dogs aged eight and over. Despite the health and welfare concerns posed by the condition, osteoarthritis o=en remains undiagnosed un,l unrelated veterinary appointments, delaying treatment, while straining the vet-owner relationship. When considering possible solutions to delayed diagnosis, human medicine demonstrates the potential for accelerometers to detect gait impairing conditions. While similar methods may be applicable to veterinary healthcare, their use would first require the identification of condition specific abnormalities in measurable gait parameters. This thesis explored potential associations between osteoarthritis, activity counts, and step count (Chapters 2 and 4), before assessing the transferral of automated methods of step detection between companion animal species (Chapter 5). Finally, the thesis established methods for deploying sensor technology in veterinary therapeutics, to address the unmet needs of vets and owners (Chapter 3). The detailed experiments highlighted associations between mean acitivity counts, osteoarthritis, and age, before exploratory analysis suggested correlations between the initial step count, age and osteoarthritis severity. Due to the capacity for mean activity counts to distinguish between sound and arthritis individuals, the measurement of this parameter may support the diagnosis of osteoarthritis. For the filtered step count, the effects of condition severity could not be isolated from those associated with age, perhaps limiting the use of this parameter to condition monitoring. Despite this, observations were based on a small cohort, and must be considered cautiously. The Delphi Consultation process then idetified piNalls in the pathway to diagnosis, before describing potenial applications for sensor technology to overcome these. A final study highlighted the accuracy of a defined canine pedometer algorithm when applied to feline data, before iden,fying areas for refinement. The studies highlight potential applicatons of sensor technology in the monitoring of canine osteoarthritis, while addressing piNalls in the pathway to effective diagnosis.