Microfluidic\nOptical Aptasensor for Small Molecules\nBased on Analyte-Tuned Growth of Gold Nanoseeds and Machine Learning-Enhanced\nSpectrum Analysis: Rapid Detection of Mycotoxins

Natural toxins, mainly small molecules,\nare a category\nof chemical\nhazards in agri-food systems that pose threats to both public health\nand food security. Current standard methods for monitoring these toxins,\npredominantly based on liquid chromatography–mass spectrometry,\nare costly, labor-intensive, and complex. This study presents the\ndevelopment of a novel microfluidic optical aptasensor for rapid detection\nof small molecules based on analyte-tuned growth of gold nanoseeds\ncombined with machine learning-enhanced spectrum analysis. We discovered\nand optimized a previously unreported growth pattern of aptamer-coated\nnanoparticles in the presence of different concentrations of analyte,\nenabling the detection of a major mycotoxin in food. The entire analysis\nwas miniaturized on a customized microfluidic platform, allowing for\nautomated spectral acquisition with precise liquid manipulation. A\nmachine learning model, based on random forest with feature engineering,\nwas developed and evaluated for spectrum analysis, significantly enhancing\nthe prediction of mycotoxin concentrations. This approach extended\nthe detection limit determined by the conventional method (∼72\nppb with high variation) to a wider range of 10 ppb to 100 ppm with\nhigh accuracy (overall mean absolute percentage error of 5.7%). The\ndeveloped analytical tool provides a promising solution for detecting\nsmall molecules and monitoring chemical hazards in agri-food systems\nand the environment.