A pilot study on real-time monitoring of canine responses to exercise

A pilot study was conducted to explore the viability of using biosensors in the field to monitor the physiology and behaviour of dogs during exercise. Seven dogs, consisting of 3 working Australian Kelpies and four companion dogs of similar size and shape, (aged 1–10 years, weight 15–22 kg) were enrolled in exercise trials. Four dogs took part in both steady-state cantering and intense incremental exercise, two dogs took part in only steady-state cantering, and one dog did not complete steady-state cantering and did not participate in intense incremental exercise. We trialed a biosensor array embedded into a commercial dog harness to measure heart rate, respiration rate, and surface temperature along with implanted data loggers (EmbediVet) to measure heart rate, temperature and motion, and ingested thermistor pills (BodyCap) to measure core body temperature. Hidden Markov models were used to organise the data into statistically distinct states of exertion. The data from all sources reliably produced a five-state solution (Akaike Information Criterion and Bayesian Information Criterion comparison of 2–7 state models), providing strong evidence that there are a stable number of states of exertion in dogs that can be identified from physiological data but not from observable behaviour such as gait, lagging, or seeking shade alone. Core body temperature peaked at 40.5°C during exercise, with heart rate strongly correlated with exercise intensity (r = 0.87, P < 0.001). Harness-embedded temperature sensors showed strong correlations with implantable sensors (r > 0.90) and thermistor pills (r > 0.85). Critically, physiological state transitions frequently occurred before observable behavioural changes, suggesting that handlers relying solely on behavioural cues, such as excessive panting, may miss early indicators of physiological stress. The harness biosensors, particularly the respiratory band, represent a viable, long-term monitoring system that can be fitted into harnesses that military working dogs currently wear, although ECG sensors required periodic gel application and EmbediVet implants showed limited feasibility due to battery life and migration risk.