Machine learning: Assisted multivariate detection and visual image matching to build broad-specificity immunosensor

The machine learning based on image matching is presented to construct a broad-specificity immunosensor for the detection of multiple ochratoxins. During the immunoreaction, ascorbic acid 2-phosphate (AAP) is catalyzed by the immobilized alkaline phosphatase to form ascorbic acid (AA), which can initiate the following reactions. First, silver ions (Ag+) can be reduced by AA to form silver coating on Au nanobipyramids (Au NBPs), changing the diameter of Au NBPs and the color of the solution. Second, AA can act as a sacrificial reagent to enhance the photoelectrochemical (PEC) current of CdS. Third, Ce4+ can be reduced to form Ce3+ and an intense fluorescence was discovered. Thus, a three-dimensional signal including colorimetry, photoelectrochemistry and fluorescence is built and then transformed to color signals for image matching. Results show that this method can predict multiple ochratoxins, suggesting that machine learning has great potential in combining multi-signal detection and multi-target detection by immunosensors.

Automated summary

A machine learning approach based on image matching is presented for constructing a broad-specificity immunosensor for detecting multiple ochratoxins. The method combines colorimetry, photoelectrochemistry, and fluorescence signals for multi-signal detection, showing great potential in predicting multiple ochratoxins. This study highlights the capability of machine learning in combining multi-signal and multi-target detection in immunosensors.