Abstract
ObjectivesThe aim of the study was to model the feeding and elimination behaviours of healthy domestic cats using data from connected devices. It was hypothesised that these behaviours would follow a regular circadian pattern.MethodsSmart feeders and connected litter boxes recording interactions (date, time, duration, quantity of faeces and urine produced, and kibble intake) were used. Data collected were processed using a Python script performing a Fourier transform to extract frequencies related to the studied behaviours and model them using sinusoidal functions. Behavioural variability was assessed by daily variation of mean (±SD) pairs of daily data and by evolution of measurements over the observation period, with 5-day moving averages.ResultsReconstructed signals from the obtained frequencies showed a pattern similar to the observed behaviours. The construction of patterns based on behaviour measurements demonstrated feeding and litter box use habits among all cats. The analysis of the variability indicated low variation over the observation period.Conclusions and relevanceModelling eating and elimination behaviours of healthy cats based on signals from connected devices was possible. Establishing such baseline behavioural profiles might help improve the detection of health and wellbeing issues in cats.