Advances in Audio Classification and Artificial Intelligence for Respiratory Health and Welfare Monitoring in Swine

Area of Science:

• Animal Science
• Biotechnology
• Artificial Intelligence

Background:

• Respiratory diseases pose significant economic and welfare challenges in modern swine production.
• Antimicrobial use is often increased due to the difficulty in early detection of swine respiratory illnesses.
• Advances in artificial intelligence (AI), specifically machine learning and deep learning, enable non-invasive monitoring through pig vocalizations.

Purpose of the Study:

• To synthesize AI-driven audio classification approaches for monitoring swine respiratory health and welfare.
• To provide a comprehensive overview of current technologies, challenges, and future opportunities in AI-based acoustic monitoring for swine.
• To serve as a reference for researchers, engineers, and industry stakeholders.

Main Methods:

• Systematic review of AI-driven audio classification in swine farming.
• Examination of biological and acoustic foundations of pig vocalizations.
• Analysis of sound acquisition, feature engineering, AI models, and evaluation methodologies.
• Discussion of commercially available systems and real-time deployment.

Main Results:

• Audio-based AI technologies show promise for early detection and continuous monitoring of swine respiratory disorders.
• Various AI models and feature engineering strategies have been applied to pig vocalizations for health assessment.
• Key challenges include data scarcity, generalization, environmental noise, and practical farm deployment.

Conclusions:

• AI-based acoustic monitoring systems offer a viable, non-invasive approach to improve swine respiratory health management.
• Future research should focus on multimodal sensing, standardized datasets, and explainable AI for enhanced system robustness and adoption.
• The integration of these technologies can lead to reduced economic losses and improved animal welfare in swine production.