Predicting heat stress index in Sasso hens using automatic linear modeling and artificial neural network

Area of Science:

• Animal Science
• Agricultural Engineering
• Environmental Physiology

Background:

• Heat stress significantly impacts poultry welfare and productivity.
• Accurate prediction of heat stress is crucial for effective management strategies.
• Robust analytical algorithms are increasingly used for forecasting environmental impacts on livestock.

Purpose of the Study:

• To forecast the heat stress index (HSI) in Sasso laying hens using analytical algorithms.
• To identify key thermo-physiological parameters influencing HSI in poultry.
• To compare the predictive performance of Automatic Linear Modeling (ALM) and Artificial Neural Network (ANN).

Main Methods:

• Utilized 167 records of thermo-physiological parameters from Sasso laying hens.
• Independent variables included housing system, age, rectal temperature (RT), pulse rate (PR), and respiratory rate (RR).
• Analyzed data using ALM (Forward Stepwise) and ANN (Multilayer Perceptron with back-propagation).

Main Results:

• Respiratory rate (RR) and pulse rate (PR) were the most important predictors of HSI.
• RR showed higher fractional importance than PR in both ALM (0.947 vs. 0.053) and ANN (0.677 vs. 0.274) models.
• Both ALM and ANN models demonstrated high accuracy in predicting HSI (r=0.980, R²=0.961 for ALM; r=0.983, R²=0.966 for ANN).

Conclusions:

• Respiratory rate is a critical indicator for predicting heat stress in laying hens.
• The developed models offer effective and accurate methods for HSI forecasting.
• Findings can inform the development of a heat stress chart based on RR for poultry management in tropical/subtropical climates.