Emerging pollutants such as tetracycline antibiotics (TCs) have garnered attention due to their ecological impacts and the evolution of drug resistance. However, the comprehensive monitoring of TCs remains inadequate. This study presents a multifunctional integrated system that addresses the limitations in current TCs research by combining detection, degradation, and bioimaging capabilities within a platform. By employing magnetic MgFe₂O₄ nanospheres, we achieve full-chain coverage for pollutant treatment, transcending the limitations of traditional single-function materials. To achieve this, we utilize machine-learning-driven precise detection technology and apply the graph neural networks model to analyze fluorescence spectra, successfully distinguishing between four TCs with highly similar structures. Subsequent dual-channel in vivo fluorescence imaging monitoring in zebrafish enables visualization of the spatial distribution of TCs types within living organisms and quantitative monitoring of TCs at different concentrations, providing a new tool for the in situ tracking of drug metabolism processes and ecotoxicological research. Lastly, the magnetic MgFe₂O₄ nanospheres exhibit efficient degradation capabilities, with degradation efficiencies exceeding 93.98% for all four TCs within 50 min. Additionally, the magnetic properties of the nanospheres address the challenge of nanomaterial recovery, enabling recycling of the catalyst. This work provides an innovative AI-material hybrid platform for intelligent pollution control.

新型污染物如四环素类抗生素（TCs）因其生态影响和药物耐药性进化而引起了广泛关注。然而，对TCs的全面监测仍然不足。本研究提出了一种多功能集成系统，该系统通过结合检测、降解和生物成像能力解决了当前TCs研究中的局限性。利用磁性MgFe₂O₄纳米球，我们实现了对污染物处理的全链覆盖，超越了传统单一功能材料的限制。为此，我们采用机器学习驱动的精确检测技术，并应用图神经网络模型分析荧光光谱，成功地区分四种结构高度相似的TCs。随后，在斑马鱼中的双通道体内荧光成像监测能够可视化活体生物中不同类型TCs的空间分布，并对不同浓度下的TCs进行定量监控，为原位跟踪药物代谢过程和生态毒理学研究提供了新的工具。最后，磁性MgFe₂O₄纳米球表现出高效的降解能力，在50分钟内所有四种TCs的降解效率均超过93.98%。此外，纳米球的磁性解决了纳米材料回收的挑战，使催化剂可以循环使用。这项工作提供了一种创新的人工智能-材料混合平台，用于智能污染控制。