Monitoring Amphetamine and Methamphetamine Mixtures Based on Deep Learning Involves Colorimetric Sensing

Precise recognition and discrimination of highly similar analytes (either in structure or property) with distinguishable sensing responses are challenging but significant in the practical application of drug seizing, food additive inspection, environmental monitoring, etc. Here, a colorimetric differentiation strategy was proposed by modulating the probe structure to influence the aggregate behaviors of the reaction products; thus, amphetamine (AMP) and methamphetamine (MA) with the sole structural difference of a methyl group were successfully discriminated. Specifically, upon recognition of the furan ring-opening reaction, the probe was screened out from a series of furan-based probes with different electron-withdrawing groups, which further facilitated the aggregate state difference of reaction products and then amplified the difference in colorimetric responses. In addition, the probe-embedded porous polymer substrate was fabricated to accelerate the response for trace AMP and MA, and the judgment of doping ratios of AMP and MA in the mixtures was realized for the first time by further combining with the self-developed Drugs Analyst as well as deep learning algorithms. Hence, we envisage that this structural-modulation-enabled colorimetric differentiation strategy will shine light on the multianalyte discrimination from aspects of optical sensing development and multidisciplinary fusion.