Validering av HOBO Pendant G data loggers förmåga att registrera beteenden och aktivitet hos suggor

Sammanfattning: There is currently no activity or behavior meters validated for sows under Swedish conditions where the sows are loose-housed during farrowing and nursing. An activity meter could in particular be used to detect sows that are getting ready for farrowing. Assistive technology in order to register different behaviors would be extremely useful in research and breeding. In this study, a HOBO Pendant G data logger was validated for two different functions. First, the function as a pure activity meter in which the accelerometer's ability to detect an upcoming farrowing was evaluated, and second, the accelerometer's ability to distinguish between the behaviors standing up, sitting down, lying ventrally and lying laterally on the right and left sides was evaluated. The accelerometer was attached to a collar around the sow's neck, and was recording its location every two seconds for six hours a day for ten days on a total of twelve sows. A total of 124 hours and 20 minutes of behavioral studies were performed in which the current behavior was recorded every 30 seconds. The results of the accelerometer's ability to predict farrowing by detecting an increase in activity in the sows on the day before farrowing was very good. Although the analysis was based on only eight farrowings, a sensitivity of 87.5% was obtained. In the case of the accelerometer's ability to distinguish between different behaviors, the result was more diverse. Lying laterally on the right and left sides could be discerned with sensitivities of 72.5 % and 83.18 %, while when the sow was sitting down could only be discerned with a sensitivity of 2.67 %. The specificity of all behaviors, however, was high, which is very useful when the aim is to rule out that a sow performed a certain behavior. In summary, there is considerable potential to use accelerometry in measuring activity of sows. To record specific behaviors may be more difficult, but with more research and improved analysis methods the sensitivities could be higher.