Integrating behavioral assessments and AI-driven diagnostics in veterinary care.

A recent collaboration between Sylvester.ai, CATalyst Council, and Austin Pets Alive! explored the use of behavioral assessments and artificial intelligence (AI) in detecting chronic pain in cats, particularly those with chronic kidney disease (CKD) and dental disease. The pilot study involved 40 mature shelter cats, with assessments via expert pain scoring, caregiver-reported behavior measures, and a computer vision-based acute pain AI model.

Elevated pain and discomfort were consistently observed in cats with chronic diseases across all three methods, confirming suspicions that these conditions contribute substantially to feline distress and highlighting a promising role for facial visual AI in chronic pain detection. The AI model showed foundational predictive capacity (balanced accuracy 53.7%) but lower performance than acute pain datasets, due to reliance on acute pain training and variable image quality. This underscores the need for future refinement and retraining to improve chronic pain detection.

Expert pain scores and caregiver-reported measures aligned reliably, suggesting these complementary approaches are effective for monitoring pain. Running and playing behaviors contributed most significantly to pain scores in cats with CKD and dental disease. Caregiver feedback supported digital tools and adding more behavioral indicators for comprehensive chronic pain monitoring.

The findings demonstrate the potential of AI in veterinary pain assessment while underscoring that larger data sets and expanded behavioral metrics are essential for progress, the collaborating organizations said. Continuous digital monitoring supports ongoing assessment rather than sporadic checks, enabling timelier treatment adjustments. The research also encourages veterinarians to more proactively screen and treat pain in high-risk populations, with AI and behavioral scoring informing relief regimens.

As AI systems are refined and retrained, diagnostic accuracy and the sophistication of pain management protocols will grow, offering a more objective and consistent approach to evaluation.

More Information:

Coleman & Slingsby, Vet. Rec. 2007
Costa et al., J. Sm. An. Prac. 2023
Dawson et al., Can. J. Vet. Res. 2017
Lascelles et al., J. Vet. Intern. Med. 2007
O’Shea & Nash, arXiv 2015
Simonyan & Zisserman, arXiv 2014.

Photo credit: istockphoto.com/Hispanolistic

Advancing Chronic Pain Detection in Cats