Machine learning-assisted ultrasensitive paper-based electrochemical sensor with aptamer-molecularly imprinted polymer dual recognition for aflatoxin B1 detection.

Aflatoxin B1 (AFB1) is a highly hazardous Group Ⅰ carcinogen that widely contaminates grain products. To realize rapid on-site detection, we developed an electrochemical sensor based on electrochemically modified graphene-carbon nanotube composite paper (EMCP), Prussian blue (PB), Au-Ag bimetallic nanoparticles and an aptamer-molecularly imprinted polymer (Apt-MIP) dual-recognition system. Density functional theory calculations were used to optimize the molecularly imprinted polymer, and machine learning was introduced to enhance the quantitative accuracy of electrochemical signals. The sensor displays a wide linear range of 10- 4-103 ng/mL with a low detection limit of 5.6 × 10- 2 pg/mL, and exhibits excellent stability, reproducibility, and selectivity. When applied to real soybean, peanut and corn samples, the recoveries ranged from 98.0% to 104.3%, matching well with HPLC results. Benefiting from low cost and good portability, this method is suitable for field detection and daily food safety supervision.