A dual-template molecularly imprinted sensor with TPE-XGBoost model for simultaneous specific detection of chloramphenicol and metronidazole in animal-derived foods

A MIP/AuNPs/GCE electrode was constructed for simultaneous detection of CAP and MNZ. • The sensor shows a wide linear range (50–1500 μM) and low LODs of 1.70 and 2.09 μM. • Wavelet transform expands the feature dimensions of the DPV signals to enhance model prediction accuracy. • TPE-optimized XGBoost realizes accurate and robust dual-analyte concentration prediction. • The method achieves satisfactory recoveries in milk, fish, and egg real samples. Food safety concerns have increased the demand for reliable simultaneous detection of chloramphenicol (CAP) and metronidazole (MNZ), as these antibiotics are strictly regulated due to their potential health risks. However, the presence of aromatic nitro groups in both compounds poses a challenge for their simultaneous detection. In this study, an electrochemical sensor integrating electrodeposited gold nanoparticles (AuNPs) with a dual-template molecularly imprinted polymer (MIP) film was developed to achieve selective recognition and signal enhancement for both analytes. However, competitive adsorption at active sites in multi-component systems leads to nonlinear signal responses, rendering traditional linear calibration ineffective. To overcome this limitation, differential pulse voltammetry (DPV) signals were transformed into two-dimensional wavelet spectrograms, from which quantitative image features were extracted, and the necessity and robustness of these features were verified using SHAP analysis. An XGBoost regression model optimized via a Tree-structured Parzen Estimator (TPE) was subsequently constructed, yielding significantly improved predictive performance. The sensor exhibited a wide linear range for both CAP and MNZ (50–1500 μM) and low detection limits (1.70 μM and 2.09 μM, respectively). Satisfactory recoveries (89.37–107.31%) were obtained in spiked milk and fish samples. The proposed TPE-XGBoost-assisted dual-template molecular sensor provides a robust and generalizable strategy for rapid and accurate multi-residue detection in complex food matrices. Download: Download high-res image (218KB) Download: Download full-size image