Real-time, non-invasive monitoring of IoT 4.0-based indicators for animal management on farms, in relation to different physiological and production stages

The ongoing digital transformation driven by Industry 4.0 has initiated a paradigm shift in animal husbandry, fostering the integration of cyber-physical systems, sensor technologies, and data analytics within livestock production systems. This dissertation investigates the deployment of Internet of Things (IoT)-enabled tools within the framework of Precision Livestock Farming (PLF), with the objective of developing and validating non-invasive, real-time monitoring systems capable of capturing physiological, morphological, and metabolic indicators in two distinct yet complementary animal models: honeybee colonies (Apis mellifera) and dairy cattle (Bos taurus). Leveraging automated three-dimensional (3D) modeling and imaging methodologies, this research elucidates the application of contactless technologies for characterizing animal health and nutritional status across varying physiological states. In Apis mellifera, morphometric analyses performed via 3D scanning revealed statistically significant divergences (p < 0.01) in thoracic width and abdominal volume between forager and nurse bees, reflecting task-specific metabolic allocation and resource utilization within the colony. The classification model achieved an accuracy of 92.5% in caste differentiation, with a 95% confidence interval ranging from 89.3% to 95.1%, thereby demonstrating the feasibility of high-resolution, in vivo phenotyping in apicultural settings. In dairy cattle, 3D body surface reconstruction techniques were integrated with metabolic biomarker profiling to refine body condition score (BCS) estimation. The automated system exhibited >93% concordance with expert visual assessment and yielded a root mean square error (RMSE) of 0.32 on a 5-point BCS scale. Model confidence was estimated at 96% (95% CI: 94.4–97.8%). Furthermore, strong correlations were observed between BCS and serum concentrations of NEFA and BHB (p < 0.001), particularly in multiparous cows during early lactation, highlighting the system’s sensitivity in detecting energy imbalance and metabolic stress. The empirical findings substantiate the potential of IoT-driven PLF approaches to enhance the precision, efficiency, and ethical sustainability of animal monitoring systems. By enabling continuous, individualized surveillance with minimal disruption, these technologies offer a robust framework for evidence-based decision-making and adaptive management. The research underscores the critical role of digitalization in advancing sustainable livestock production and affirms the necessity of cross-disciplinary integration to address the multifactorial challenges of modern animal agriculture.