AI-driven Cu/Mn-CeO₂ nanozyme-functionalized paper-based biosensor for quantitative monitoring of alternariol in food

The lack of rapid, user-friendly methods for alternariol (AOH), a prevalent emerging mycotoxin, presents a considerable analytical challenge. This work introduces an integrated sensing strategy advancing from classical homogeneous liquid-phase detection to a portable AI-enhanced paper-based biosensor. The system employed a Cu/Mn-CeO₂ nanozyme as signal generator and amplifier, and exploited the inhibitory effect of AOH on acetylcholinesterase (AChE) to suppress the thiocholine (TCh)-mediated reduction of oxidized 3,3',5,5'-tetramethylbenzidine (oxTMB), enabling quantitative colorimetric readout. It achieved ultra-sensitive detection in the liquid phase with a limit of detection (LOD) of 0.016 pg/mL. By incorporating an aptamer affinity column for sample cleanup, the assay was transferred to a foldable paper platform. An AI-driven Monte Carlo color analysis method addressed issues of color heterogeneity and subjective interpretation, yielding an LOD of 0.12 μg/kg in wheat. This study provides a point-of-need solution for AOH screening and establishes a generalizable framework for detecting contaminants in complex samples.