Skip to main content
ARS Home » Plains Area » Clay Center, Nebraska » U.S. Meat Animal Research Center » Nutrition, Growth and Physiology » Research » Publications at this Location » Publication #354896
Research Project: Improve Nutrient Management and Efficiency of Beef Cattle and Swine

Location: Nutrition, Growth and Physiology

Title: Identifying anomalous decreases in feeding time of grow-finish pigs

Author
item ADRION, FELIX - University Of Hohenheim
item Brown-Brandl, Tami
item JONES, D - University Of Nebraska
item GALLMANN, E - University Of Hohenheim

Submitted to: Meeting Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 7/13/2018
Publication Date: 8/13/2018
Citation: Adrion, F., Brown-Brandl, T.M., Jones, D., Gallmann, E. 2018. Identifying anomalous decreases in feeding time of grow-finish pigs. Proceedings of the 10th International Livestock and Environment Symposium (ILES), September 25-27, 2018, Omaha, NE. Paper No. ILES 18-042. https://doi.org/10.13031/iles.18-042.
DOI: https://doi.org/10.13031/iles.18-042

Interpretive Summary:

Technical Abstract: Disruptions in feeding behavior can be indicative of a beginning illness and other impairments of animal well-being. The objectives of this study were to 1) develop an autoregressive linear model to predict daily feeding time of grow-finish pigs, 2) detect anomalous decreases in daily feeding time, and 3) compare the algorithm with animal caretaker records. The daily feeding time was collected with a low-frequency RFID system for a total of 2880 pigs over 12 different groups. Animal caretakers checked the pigs for health daily. Detected illness and subsequent treatments were recorded. Numerical computing software was used to develop an autoregressive linear model using an expanding and moving time window to predict daily time at the feeder for each pig. In a first step, the model and an algorithm using a z score threshold to detect large decreases were calibrated using caretakers’ detections of pneumonia that were associated with severe drops in feeding time. The health warnings of the final model and algorithm were compared to the caretaker diagnoses in the reference feeding period. The algorithm detected 62 % of severe drops related to an illness event. However, the algorithm also resulted in a high number of false negatives (illness not detected) and false positives (alarm for a healthy animal). These results can be partly explained with the lack of a reliable and repeatable gold standard, but also reveal the potential of such a warning system to assist the animal caretaker to prevent unnecessary treatments of animals and also detect more illnesses.