Explainable Artificial Intelligence (XAI) Applied to Genomic Predictive Modeling of Feeding Behavior in Pigs

• Grant number: 25/10503-7
• Support Opportunities: Scholarships in Brazil - Post-Doctoral
• Start date: September 01, 2025
• End date: August 31, 2028
• Field of knowledge: Agronomical Sciences - Animal Husbandry - Genetics and Improvement of Domestic Animals
• Principal Investigator: Aline Silva Mello Cesar
• Grantee: Izally Carvalho Gervásio
• Host Institution: Escola Superior de Agricultura Luiz de Queiroz (ESALQ). Universidade de São Paulo (USP). Piracicaba , SP, Brazil

Abstract

This project aims to apply explainable artificial intelligence (XAI) techniques to the genomic prediction of feeding behavior traits in Yorkshire and Landrace pigs. Previously available genotyping data will be used, associated with phenotypic records of daily visit duration, feed intake per feeder visit, and number of feeder visits per day per animal. To develop accurate predictive models, these data will be analyzed using machine learning algorithms such as XGBoost, Random Forest, and Neural Networks. The innovative aspect of this proposal lies in the integration of model interpretability tools, including SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations), allowing for the identification of genetic markers (single nucleotide polymorphisms - SNPs) with the greatest influence on the studied phenotypes. In addition to improving the predictive power of models used in breeding programs, the project aims to provide biologically meaningful insights, fostering more efficient and interpretable genomic selection with practical applicability. Including multiple commercial breeds with more than 50,000 genotyped animals is expected to increase the robustness and generalization potential of the studied approach. To date, no studies have integrated XAI with genomic prediction of feeding behavior and performance-related traits in swine production, making this proposal innovative and unprecedented, with strong potential impact on animal science and the livestock industry. (AU)