Design of an Intelligent Profiling Platform for Multiple Antibiotic Discrimination Enabled by Time-Resolved...

Nanozymes with peroxidase-like (POD) activity have broad applications in biosensing, but their catalytic activity is often optimal under acidic conditions, which severely limits their applications in neutral environments. Here, we constructed a histidine-functionalized cupric oxide (CuO@His) nanozyme via defect engineering. Relative to CuO, CuO@His exhibited enhanced POD activity at neutral pH and was capable of catalyzing 3,3',5,5'-tetramethylbenzidine (TMB) to generate blue oxidized TMB, which showed absorption peaks at 370 and 652 nm. Subsequently, we employed the recursive feature elimination strategy to screen the optimal wavelength-time combinations as the sensing channels and further established the characteristic "fingerprints" of different antibiotics. Moreover, with the assistance of machine learning algorithms, a concentration- and matrix-independent antibiotic identification model was established under neutral conditions. Finally, a visual intelligent sensing platform based on deep learning was constructed. This work provides a new strategy for the efficient identification of multiple antibiotics at neutral pH.