CompanionCare Analytics Model for Assessing Domestic Pet Behavior and Environmental Wellbeing

Abstract

The increasing popularity of wearable devices for monitoring pets has opened up possibilities to go beyond activity metrics and towards a holistic view of pet wellness. In this paper, it introduce CompanionCare, which is an analytics system based on integration of wearable sensory measurements, household environmental measurements, and behavior detection in order to measure the behavior and environment-related wellbeing of pets. Contrary to existing systems focused solely on the activity classifications, CompanionCare uses a triple-layered approach consisting of data collection, behavior analysis, and wellbeing measurement, and converts the collected data from accelerometers, temperature, humidity, noise, and light exposure to Environmental Wellbeing Index (EWI). Using the supervised classifiers, CompanionCare detects core behavior classes such as resting, walking, running, eating, scratching, and vocalizing, and correlates changes in behaviors with environmental factors such as temperature changes, increased noise levels, and unusual light exposure. A comparative study conducted based on benchmarks provided in wearable sensor-based animal behavior literature demonstrates the compatibility of the introduced pipeline not only with competitive behavior classification accuracy, but also with continuous wellbeing assessment capability. These findings imply that the inclusion of environmental information in addition to behavior tags can help detect subtle stress incidents that are usually missed by the activity recognition systems. The paper also examines issues such as sensor positioning, data security, and power consumption related to implementing the solution in a home environment. CompanionCare is presented as a decision support system rather than a diagnostic one, designed to aid rather than substitute veterinary assessment of a pet’s condition.