Published in Animals, this study by Seon-Chil Kim and Sanghyun Kim addresses a critical challenge in companion animal care: most dog owners lack veterinary training, making it difficult to detect early signs of illness. The researchers developed an AI disease prediction algorithm capable of translating everyday behaviors into a clear, numerically based Health Score to support timely decision-making.
To build the model, dogs were equipped with sensor-fitted leashes that measured a range of daily activities, including scratching, licking, swallowing, and sleep patterns. Three months of baseline behavior were used to train the AI to recognize abnormalities, while nine months of continuous sensor monitoring validated the system in real-world conditions.
The resulting Health Score categorizes canine well-being into ten levels, with higher scores indicating better health. Dogs scoring below 5 are flagged for veterinary consultation, while those scoring above 5 are considered healthy. When compared against professional veterinary assessments, the model achieved an 87.5% concordance rate, demonstrating strong reliability.
This research shows how integrating behavioral analytics, sensor technology, and AI-driven prediction can offer dog owners a practical, accessible tool for monitoring health. As sensor precision and behavioral analysis methods continue to improve, such systems may transform preventive care and support earlier detection of disease in companion dogs.
Source: Kim, S.-C., & Kim, S. (2024). Development of a Dog Health Score Using an Artificial Intelligence Disease Prediction Algorithm Based on Multifaceted Data. Animals. Published January 1, 2024.
