Potential for real-time health and welfare monitoring in experimental rabies infection in red fox (Vulpes vulpes) using implants

Monitoring and improving animal health and welfare is a pressing need. Technological advances are increasing our ability to do so in a range of settings but currently, there are limited uses in experimental settings for less common model animals. Red foxes (Vulpes vulpes) are maintained by the French National Agency for Food, Environmental, and Occupational Health and Safety (ANSES) for regulatory and research purposes, primarily related to its role as a European Union and French Reference Laboratory for rabies and Echinococcus spp. In 2022, four foxes were surgically fitted with internal implants that recorded body temperature and activity levels and subsequently inoculated with rabies virus (RABV). Retrospectively, we tested the potential of these two variables to provide an early warning for the onset of rabies symptoms. We applied two anomaly detection algorithms (Shewhart and EWMA) with varying confidence levels to the data sets and compared how early the different models could detect significant changes while limiting false alarms during the calibration period. We hypothesized that body temperatures would rise significantly and foxes would significantly alter their activity levels at the beginning of infection, both at an earlier stage than is detectable through direct observation. We found that foxes significantly changed their activity during infection. We were best able to detect these changes using the EWMA algorithm, in some cases producing consecutive alarms up to two weeks before the death of an animal, while limiting false alarms. We found no evidence of fever in any of the infected foxes and body temperature did not appear to be a reliable indicator of foxes health. While here we applied our methods to a particularly severe and rarely implemented model with a very small sample size, this proof of concept illustrates the potential of these methods for a wide range of other situations that would benefit from similar long-term monitoring, including experimental protocols with milder clinical signs and routine monitoring for unexpected declines in health or welfare.