ARTIFICIAL INTELLIGENCE AND DEEP LEARNING FOR AUTOMATED DIAGNOSIS OF BOVINE RESPIRATORY DISEASE COMPLEX FROM INFRARED THERMOGRAPHY AND AUDIO SIGNAL

Abstract

Bovine Respiratory Disease Complex (BRD) remains a persistent challenge in cattle health management, necessitating early, accurate, and scalable diagnostic solutions. This study presents a novel AI-enabled framework that integrates infrared thermography and respiratory audio signals to facilitate real-time BRD detection using deep learning algorithms. A hybrid CNN-LSTM architecture was trained on multimodal datasets, capturing subtle patterns in thermal gradients and acoustic anomalies. Results were validated across nine detailed tables containing physiological and behavioral metrics for over 180 cattle instances, demonstrating a consistent correlation between elevated temperature, increased cough frequency, and BRD risk. Twelve complex visualizations—including line, bar, scatter, and hybrid plots—illustrated the efficacy of fused features in enhancing diagnostic accuracy. The model achieved high sensitivity and specificity in BRD classification and proved effective in real-world deployment, offering sub-second diagnostic latency. These findings underscore the potential of AI-powered tools in non-invasive veterinary diagnostics, providing farmers with actionable insights and early intervention capabilities. This integrated approach advances the field of precision livestock farming and opens new pathways for scalable, automated animal health monitoring systems.