Machine learning-assisted self-powered ear tag for animal welfare

Metabolic health serves as a crucial indicator of animal welfare, yet nutritional imbalances in intensive farming diets often induce metabolic dysregulation. ‌However, precise identification of metabolically abnormal animals within intensive production systems remains challenging. Here we report a machine learning-assisted self-powered ear tag that enables streamlined large-scale deployment in livestock production, providing continuous monitoring of ion homeostasis. The ear tag is powered by a hybrid energy harvesting module based on a triboelectric nanogenerator and a solar cell, sustaining energy-autonomous operation through an optimized duty-cycled strategy. Leveraging a microneedle-based multiplexed biosensing module, the system facilitates minimally invasive and time-resolved monitoring of pH, K+, and Ca2+ fluctuations in interstitial fluid. By implementing a machine learning pipeline to decode the coordinated dynamics of these multi-ionic markers, the platform effectively distinguishes among five distinct welfare-related states with an average classification accuracy over 95%. This performance is rigorously validated via leave-one-animal-out cross-validation across 5399 sampling windows from three independent animals. By bridging energy-autonomous operation with intelligent metabolic profiling capabilities, the self-powered ear tag establishes a scalable technological paradigm for next-generation precision livestock health monitoring and welfare management. Authors combine triboelectric and solar energy to fabricate machine learning-assisted, self-powered ear tags for livestock that continuously monitor pH, potassium, and calcium ions using minimally invasive microneedle biosensors. The system achieves over 95% accuracy in classifying up to 5 metabolic states.