A team of AI scientists and seismologists has developed a pioneering self-supervised framework, DASFormer, for high-resolution earthquake monitoring. Built on a two-stage, coarse-to-fine architecture, DASFormer is pre-trained on vast amounts of unlabeled Distributed Acoustic Sensing (DAS) data collected from existing fiber-optic cables (e.g., Internet cables). Acting like a “self-taught seismologist”, the model first learns the predictable patterns of background noise and then flags earthquake signals as anomalies that defy its forecasts. This novel approach demonstrates superior performance, outperforming other state-of-the-art models. Its versatility extends to challenging environments such as the seafloor, underscoring its potential for scalable, automated seismic intelligence.

Flexible composites-based piezoelectric nanogenerator (PENG) with low cost, stable properties and sensitivity to mechanical deformation is highly suitable to construct self-powered sensing layer for distributed electrical transmission power lines, and this innovation can help reduce manual maintenance costs. However, the lower output performance of the PENG hinders its integration with energy management circuits and signal recognition systems. In this study, a high-performance PENG was achieved by designing branch-heterostructure piezoelectric ceramic fibers, which can enhance the charge transport mechanisms and induced polarization. Moreover, an intelligent Power Internet of Things system through the synergistic integration of this high-performance PENG and learning-assisted data analytics has been constructed, which enables accurate self-powered real-time monitoring of abnormal vibration states in transmission power lines with approximately 96% identification accuracy. This work not only provides an effective strategy to enhance PENG performance, but also offers a solution to improve the reliability of power grid operations and optimize maintenance efficiency. Recently, a team of material scientists led by Haowei Lu from Henan University, China, prepared high-performance PENG based on a new piezoelectric ceramic fiber, which is beneficial to the improvement of electrical output performance by enhanced induced polarization and directed charge transport mechanism. Moreover, based on this PENG, a power grid transmission line vibration determination system with high identification accuracy can be constructed in this study.

The gut microbiota is widely recognized as a central regulator of human health and disease. Medicine-food homologous resources, leveraging their inherent safety and multi-target characteristics, serve as pivotal modulators for intervening in metabolic, inflammatory, and immune-related disorders via microbiota regulation. However, the inherent complexity, substantial interindividual variability, and dynamic nature of the gut microbiome remain major hurdles to achieving precise interventions. This perspective delineates a novel paradigm for precision medicine-food intervention, built upon three interconnected and cutting-edge directions: (1) targeting key microbial metabolites, (2) advancing targeted delivery technologies for beneficial microbes, and (3) implementing artificial intelligence (AI)-assisted personalized microbiome functional profiling. This triad synergistically addresses the challenge of individual variability and paves the way for highly effective and precise interventions.

Antimicrobial resistance has become one of the top global public health and development threats due to the misuse and overuse of antimicrobials in humans, animals, and plants. Researchers are leveraging artificial intelligence and interdisciplinary approaches to design antimicrobial peptides (AMPs) that show a reduced risk of inducing resistance. Precise targeting design makes AMPs more efficient for combating drug-resistant bacteria and fungi, with applications spanning medicine, agriculture, and food safety.