Kinetic-based supervised opioid discrimination using unmodified AgNPs as colorimetric sensor array

Pattern recognition using the nanoparticle colorimetric sensor arrays is a powerful analytical tool for detection and discrimination of analytes. In this study, the unmodified silver nanoparticles (AgNPs) with different diameters were employed for discrimination of three opioid drugs methadone, morphine, and tramadol based on the difference in the reaction kinetics between the drug and AgNPs. Discrete wavelet transform with various mother wavelets was used to preprocess the obtained big data and reduce their dimensionality, and then neural network-based supervised pattern recognition methods like Counter Propagation Artificial Neural Networks (CPANN), Supervised Kohonen Networks (SKN), and XY-fused (X-YF) networks were used for discrimination after optimizing the network parameters with genetic algorithm. The best pattern recognition performance was achieved using the SKN and Daubechies (Db) mother wavelet with an accuracy of 98% in discrimination of the analytes in spiked real serum samples.