Wearable sensor shown to specifically quantify pruritic behaviors in dogs

Joel D. Griffies, Jason Zutty, Marcel Sarzen, Stuart Soorholtz

Wearable technology is an exciting new field in humans and animals. In dogs activity monitors have helped to provide objective measurement tools where pet owner observation had been the only source of information. Previous research has focused on measuring overall activity versus rest. This has been relatively useful in determining changes in activity in orthopedic disease or post-surgical cases [Malek et al., BMC Vet Res 8:185, 2012, Yashari et al., BMC Vet Res 11:146, 2015]. Assessment of pruritus via changes in activity, however, requires an assumption that increased activity is due to scratching or other pruritic behaviors. This is an inaccurate method with obvious flaws as other behaviors may also register as greater activity. The objective of this study was to validate the ability of a multidimensional high frequency sensor and advanced computer analysis system, (Vetrax®, AgLogica Holdings, Inc., Norcross, GA, USA) to specifically identify pruritic behaviors (scratching and head shaking). To establish differences between behaviors, sensor and time stamped video data were collected from 361 normal and pruritic dogs. Video annotations were made by two observers independently, while blinded to sensor data, and then evaluated for agreement. Annotations that agreed between the two were used for further analysis. The annotations specified behaviors at specific times in order to compare with sensor data. A computer algorithm was developed to interpret and differentiate between these behaviors. Test subject data was then utilized to test and score the system's ability to accurately predict behaviors. Results for prediction of head shaking behavior included sensitivity and specificity of 72.16% and 99.78% respectively. Analysis of scratching produced sensitivity and specificity of 76.85% and 99.73% respectively. These results illustrate the ability of the system to accurately report both scratching and head shaking with an overall accuracy of 99.24% and 99.56% respectively. This study validates the use of this system to accurately and objectively report scratching and head shaking in dogs. While a small portion of scratching or head shaking behaviors may be missed, as indicated by the sensitivity, when detected, the confidence that these behaviors occurred is extremely high. These factors make this system a very useful tool for objective assessment of pruritus in clinical and research settings.