Dual-Mode Colorimetric/SERS Lateral Flow Immunoassay with Machine Learning-Driven Optimization for Ultrasensitive Mycotoxin Detection.

Abstract

Detecting and quantifying mycotoxins using LFIA are challenging due to the need for high sensitivity and accuracy. To address this, a dual-mode colorimetric-SERS LFIA was developed for detecting deoxynivalenol (DON). Rhodium nanocores provided strong plasmonic properties as the SERS substrate, while silver nanoparticles created electromagnetic "hotspots" to enhance signal sensitivity. Finite element modeling optimized the electromagnetic field intensity, and Prussian blue generated a distinct signal at 2156 cm^-1, effectively reducing background interference. This dual-mode LFIA achieved a detection limit of 4.21 pg/mL, 37 times lower than that of colloidal gold-based LFIA (0.156 ng/mL). Machine learning algorithms, including ANN and KNN, enabled precise classification and quantification of contamination, achieving 98.8% classification accuracy and an MSE of 0.57. These results underscore the platform's potential for analyzing harmful substances in complex matrices and demonstrate the important role of machine learning-enhanced nanosensors in advancing detection technologies.