Integrating Deep Learning and Transcriptomics to Assess Livestock Aggression: A Scoping Review.

Area of Science:

• Animal Science
• Bioinformatics
• Artificial Intelligence

Background:

• Livestock aggression poses significant challenges to animal welfare, farm safety, economic viability, and selective breeding programs.
• Emerging technologies like deep learning-based video analysis and transcriptomic profiling offer novel avenues for understanding and monitoring aggressive behaviors.

Purpose of the Study:

• To conduct a scoping review on the application of deep learning video analysis and transcriptomic profiling in livestock aggression research.
• To identify current trends, research gaps, and future directions for these innovative methodologies.

Main Methods:

• A comprehensive literature search was conducted across PubMed, Scopus, and Web of Science databases for studies published between 2014 and April 2025.
• The review analyzed 268 original studies, categorizing them into AI-driven behavioral phenotyping (250 papers) and transcriptomic investigations (18 papers).

Main Results:

• The majority of studies focused on pigs and cattle, with limited attention given to poultry, small ruminants, camels, and fish.
• Key advancements include Convolutional Neural Network (CNN)-based systems for object detection and pose estimation, and transcriptomic identification of molecular pathways related to aggression and stress.
• A significant gap exists in studies combining both video analysis and transcriptomic approaches.

Conclusions:

• Inconsistent behavioral annotation, insufficient real-farm validation, and limited cross-modal integration are major barriers to progress.
• Standardized behavioral definitions, multimodal datasets, and integrated pipelines linking phenotypic and molecular data are proposed.
• These innovations are crucial for advancing livestock welfare, precision breeding, and sustainable animal production.