Machine learning-aided quantitative detection of two mixed antibiotics in a narrow concentration range based on oxygen incorporation-induced 3D SERS substrates

Amoxicillin (AMO) and ciprofloxacin (CIP) as frequently-used antibiotics are often simultaneously taken for anti- infection to protect pig and poultry production, while their effective therapeutic window is narrow. How to quantitatively detect the two antibiotics mixed in a narrow concentration range is essential to ensure their treatment efficacy and food safety. In this study, quantitative detection of AMO and CIP mixed in animal flesh in one order of magnitude concentration range was performed using a triple SERS signal amplification sensor combined with machine learning. The SERS sensor based on oxygen incorporation (Oi)-induced 3D NiO nano- flowers (NFs) coated with Ag nanoparticles included 3D hot spot effect from strong internal electromagnetic couple in NiO NFs, the Oi effect of NiO and synergistic charge transfer effect of between NiO and Ag. Based on triple enhancement effects, sensitive SERS detection of AMO and CIP was made with the limits of detection of 1 nM and 100 nM. Through the use of machine learning techniques like principal component analysis and partial least square regression, qualitative and quantitative detection of AMO and CIP in their mixture from 1.33 mu M to 8.69 mu M was successfully achieved, where the average proportion of absolute prediction error (p <= 0.15) for concentration was 98.06 %.