RESUMO
Low mass transfer efficiency and unavoidable matrix effects seriously limit the development of rapid and accurate determination of biosensing systems. Herein, we have successfully constructed an ultra-rapid nanoconfinement-enhanced fluorescence clinical detection platform based on machine learning (ML) and DNA xerogel "probe", which was performed by detecting neutrophil gelatinase-associated lipocalin (NGAL, protein biomarker of acute kidney injury). By regulating pore sizes of the xerogels, the transfer of NGAL in xerogels can approximate that in homogeneous solution. Due to electrostatic attraction of the pore entrances, NGAL rapidly enriches on the surface and inside the xerogels. The reaction rate of NGAL and aptamer cross-linked in xerogels is also accelerated because of the nanoconfinement effect-induced increasing reactant concentration and the enhanced affinity constant KD between reactants, which can be promoted by â¼667-fold than that in bulk solution, thus achieving ultra-rapid detection (ca. 5 min) of human urine. The platform could realize one-step detection without sample pretreatments due to the antiligand exchange effect on the surface of N-doped carbon quantum dots (N-CQDs) in xerogels, in which ligand exchange between -COOH and underlying interfering ions in urine will be inhibited due to higher adsorption energy of -COOH on the N-CQD surface relative to the interfering ions. Based on the ML-extended program, the real-time analysis of the urine fluorescence spectra can be completed within 2 s. Interestingly, by changing DNA, aptamer sequences, or xerogel fluorescence intensities, the detection platform can be customized for targeted diseases.