Sensor Fusion for Predicting Impaired Resilience in Pigs: Towards an Early Warn-ing System

Concerns about the health and welfare of intensively farmed pigs have raised questions about their resilience in coping with environmental stressors. This study investigates the potential of various sensors and their combination in predicting factors contributing to reduced resilience in pigs, aiming to develop an early warning system. In our experiment, sensors were placed in a pig farm containing 12 pens, each pen with thirty weaned piglets. Piglets were introduced post-weaning, staying until reaching approximately 25 kilograms live weight and data will be collected during at least 4 subsequent production cycles. Sensors included water-intake sensors, climate sensors (CO2, NH3, temperature, illumi-nance), audio recording sensors for coughing, and video cameras. Additional data were manually collected to determine when impaired resilience occurred, like tail and ear-damage. We conducted a MANOVA test to determine how different environmental con-ditions can result in biting behaviour among animals. Our initial findings indicate no sta-tistically significant relationship, potentially due to the limited sample size. Future efforts will involve incorporating more diverse sensor data to enhance the robustness of our anal-ysis.