Early detection of individual growing pigs’ sanitary challenges using functional data analysis of real-time feed intake patterns

Abstract This article is concerned with the conception of an automatic numerical procedure which, integrated into automatic feeders, can identify changes in the feed intake patterns of individual pigs, thus allowing early detection of potential sanitary challenges within the herd. More precisely, the proposed numerical procedure analyzes every day, and for each pig within the herd, feed intake data collected during 5 consecutive days (memory lag) to predict the feeding patterns of the following day. Then, the procedure evaluates, for each animal, the difference between the predicted and the observed feeding patterns and automatically detects if this difference is greater than a given threshold. In this case, a signal is sent to a monitoring center and the animal can be placed under observation.