Evolution from Analyte to Sensitive Signal Probe: Norfloxacin-Based Assembly for Machine-Learning-Assisted Discrimination of Phosphates and Monitoring of ATP Hydrolysis

The development of sensor arrays for effective discrimination of structurally similar physiological phosphates (PPs) in mixtures presents a considerable challenge. We propose a new strategy based on concentration-regulated supramolecular assembly to construct a sensor array, integrated with machine learning, for the identification of PPs and real-time monitoring of ATP hydrolysis. Norfloxacin (NF) is upgraded into a signaling element that forms four dynamic NF-Eu3+ complexes serving as a four-element sensor array. Gradient fluorescence quenching is generated during NF-Eu3+ assembly at varying ratios; subsequent PP introduction modulates emission via competitive Eu3+ sequestration and/or ternary NF-Eu3+-PP complexation driven by π–π stacking interactions, which results in either fluorescence recovery or further quenching. The developed machine learning-assisted array accurately distinguishes five PPs and achieves ultrasensitive ATP detection with a limit of detection as low as 0.41 nM, while exhibiting high selectivity and strong anti-interference capability. Its practical utility is demonstrated in real-time monitoring of ATP hydrolysis. A smartphone-integrated portable system was also developed for on-site rapid detection of phosphates in serum. This study not only showcases a new approach to resource utilization of norfloxacin in sensing applications but also provides a strategy for developing efficient and low-cost platforms for phosphate detection.