Machine learning-assisted colorimetric sensor array on a superwettable microchip for on-site authentication of ginseng.

Panax ginseng is widely used for its significant bioactivities, but adulteration and quality counterfeiting remain critical issues. Herein, a machine learning-assisted superwettable colorimetric sensor array microchip is proposed for the rapid ginseng authentication. The microchip targets diol groups of ginseng-derived monosaccharides using 3-nitrophenylboronic acid as the cross-reactive receptor, generating distinct colorimetric fingerprints via pH response and indicator displacement assay (IDA) mechanisms. The excellent determination capability of the microchip was demonstrated through the classification and semi-quantitative analysis of monosaccharides in ginseng hydrolysates. Coupled with advanced machine learning algorithms, high-accuracy analysis was achieved for ginseng species identification and authenticity classification, reaching 96.49% and 99.02%, respectively. Furthermore, support vector regression (SVR) enabled precise quantitative analysis of adulteration ratios with excellent predictive performance (R = 0.9869). The proposed method was successfully applied to identify commercial ginseng samples with reliable results. This work provides a low-cost, rapid, and instrument-free strategy for on-site quality control of ginseng products.