Web system-assisted ratiometric fluorescent probe embedded with machine learning for intelligent detection of pefloxacin

It is urgent to realize the quantitative detection of pefloxacin (PFLX), which has been misused to accumulate in the environment and animals. The traditional detection method is not only time-consuming and laborious, but also can only detect a single sample at a time. In this research, a BSA-AuNCs-GMP-Tb ratiometric fluorescent nanocomposite probe with red fluorescence in the initial state was constructed by in situ preparation of guanine nucleotide-terbium ion (GMP-Tb) complex using bovine serum albumin-coated gold nanoclusters (BSA-AuNCs) as the substrate material. Under the optimal experimental conditions, the nanoprobe had a wide visual detection range (0.01 mu M-60 mu M) with an ultra-low detection limit (5.37 nM) for PFLX. A paper-based visualized sensor was further designed with a smartphone color recognition app to achieve qualitative and semi-quantitative detection of PFLX. In addition, merging You Only Look Once version 5 (YOLO v5) target detection algorithm and machine learning regression algorithm, a web-based PFLX concentration prediction system was developed to achieve the quantitative detection of PFLX in actual environment samples. The system was developed to achieve batch detection of PFLX in environmental samples, which greatly improved the speed of PFLX detection.