Enhancing Detection of Feline Chronic Kidney Disease Through Smart Litter Box Monitoring

Abstract

Chronic kidney disease (CKD) is a prevalent condition in cats and recognized as a leading cause of mortality among cats generally, yet it is difficult to detect by cat caregivers, especially in its early stages. This retrospective study aimed to enhance detection of CKD by cat caregivers through a smart litter box monitor technology that tracks feline elimination behaviors. We established cohorts of cats with CKD and cats with no known conditions and analyzed behavioral features captured by the monitor. A mixed-effects model identified significant differences, which were then integrated into a machine learning framework for predictive modeling. The model achieved a weighted F1-score of 92.7% in training using cross-validation and 89.9% in validation, demonstrating high precision for CKD predictions. Key findings included a behavioral profile unique to cats with CKD characterized by increased urination frequency, longer elimination durations, and reduced post-elimination covering behavior. These results suggest that the smart litter box monitor can provide valuable, continuous, and non-invasive data for CKD detection and management.