The Cobb R&D team has embarked on a groundbreaking collaboration with researchers from the University of Georgia and the University of Wisconsin-Madison, aiming to leverage feeding behavior as a predictor of broiler chicken health. The foundation of this study rests on a well-established observation: animals, including chickens, alter their eating habits, social interactions, and general activity when they experience illness or injury.

Experimental Design and Methodology

To study these behavioral changes, Cobb R&D and its academic partners designed an innovative experiment that tracked individual chickens using radio-frequency identification (RFID) technology. The birds were fitted with RFID tags affixed to their wings, while specialized feeding stations equipped with antennas were developed.

These feeding stations were meticulously engineered by Cobb’s team to ensure that only one bird could access food at a time, allowing for accurate tracking of individual feeding behaviors. Over a span of five years, the study monitored more than 95,000 broiler chickens and recorded data on nearly 100 million feeder visits.

Data Collection and Analysis

The data collected from these visits was extensive, summarizing various feeding traits such as:

• Feed intake
• Number of visits
• Time spent at the feeder
• Time intervals between feeder visits
• Number of feeders accessed

Moreover, the feeding stations were equipped with scales to measure feeding amounts (g/meal) and feeding rates (g/hour), facilitating comprehensive data recording. A schematic representation of the experimental design illustrates how the RFID transponder activates during feeder visits, transmitting data that is subsequently decoded and stored in a cloud database.

Machine Learning Implementation

To analyze the extensive data collected, the team trained and tested five distinct machine learning models, each employing a unique classification system to uncover subtle patterns that could indicate illness in the birds. Notably, two algorithms—Gradient Boosting Machines (GBM) and Support Vector Machines (SVM)—demonstrated superior performance, achieving mortality predictions as much as one to three days ahead of actual events.

Implications for Broiler Chicken Management

This innovative monitoring system presents several significant benefits for poultry farming:

• Early Intervention: By predicting health issues in advance, farmers can intervene promptly to treat ill or injured birds, greatly enhancing animal welfare.
• Economic Savings: Reducing mortality rates can lead to substantial financial benefits for poultry producers.
• Increased Biosecurity: Automated monitoring minimizes the necessity for in-house flock inspections, aiding in maintaining biosecurity standards.
• Data-Driven Management: The collected data can be analyzed to inform future management decisions, leading to improved precision and outcomes.