Defect-Engineering-Tailored Triple-Enzyme-Mimetic Heterojunction Nanozyme-Based Colorimetric Sensor Array for AI-Powered On-Site Intelligent Assay of Multiple Antibiotic Residues

Maintaining high levels of multienzyme-like activities within a single nanoparticle-based sensing array to boost the efficient simultaneous on-site assay of multiple antibiotic residues at low regulatory threshold levels is critical for the “One Health” initiative, but remains an ongoing challenge. Herein, a defect-engineered Pt−CuO/Cu2O heterojunction nanozyme was rationally designed with enhanced peroxidase- (POD), oxidase- (OXD), and laccase-like (LAC) activities, in which each component performs its designated function with efficient coordination. The introduction of Pt sites and increased oxygen vacancy (OV) through defect engineering modulate the heterointerface electronic structure and optimize the d-band center, as corroborated by density functional theory calculations. Such electronic modulation enhances the adsorption energy of Pt−CuO/Cu2O towards the substrates such as H2O2 (−1.02 eV to −1.49 eV) and O2 (−0.76 eV to −1.12 eV), thus boosting its triple-enzyme-like catalytic performance concurrently. Leveraging the distinct modulatory effects of β-lactams (BLs), tetracyclines (TCs), and fluoroquinolones (FQs) on these three enzyme-like activities, a single-receptor, three-channel colorimetric sensing array was constructed. Coupled with a home-made, highly-integrated and portable device for RGB signal acquisition and machine learning-assisted classification-regression models, the array achieved accurate on-site intelligent analysis of the three types of antibiotics across 0.2 μM to 100.0 μM within only 15.0 min, with performance approaching and even exceeding the regulatory limits. Performing the on-site assay on spiked milk powder samples produced satisfactory accuracy and recoveries, which were validated against authoritative LC−MS. This work establishes a defect-engineering paradigm for maintaining triple-enzyme-mimetic activities at high catalytic levels in a single nanoparticle-based efficient sensing array, enabling the on-site sensitive and accurate discrimination of multiple antibiotic residues at low concentrations through a machine learning-assisted colorimetric sensing array, thus offering both theoretical guidance and a deployable solution for food safety monitoring of multiple antibiotic residues.