SERS combined with self-optimizing machine learning algorithm for quantitative detection of norfloxacin and ciprofloxacin in milk

Silver nanoparticles prepared by trisodium citrate system have the advantages of low cost, simple preparation and excellent SERS performance, which are widely used in food safety monitoring. However, the uncontrollable size limits the practical application of silver nanoparticle SERS substrates. In this work, we optimized the size of silver nanoparticles by controlling the reduction rate of silver ions through the injection speed of silver nitrate. The results indicated that an injection rate of 2 mL/min produced silver nanoparticles with the most uniform SERS performance, achieving an RSD of Raman spectral intensity only 5.07 %. This silver nanoparticle-based SERS substrate was then employed to detect ciprofloxacin and norfloxacin in milk, with a detection limit of 10 ppb. To accurately quantify the concentrations of ciprofloxacin and norfloxacin, a self-optimizing concentration prediction model was established. The findings demonstrated that silver nanoparticles with stable SERS performance, combined with a self-optimizing machine learning algorithm, could rapidly and accurately identify the concentrations of ciprofloxacin and norfloxacin in milk, with a coefficient of determination between predicted and actual concentrations as high as 0.996.