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

Area of Science:

• Analytical Chemistry
• Nanoscience
• Biomedical Engineering

Background:

• Mycotoxin detection via lateral flow immunoassay (LFIA) faces challenges in achieving high sensitivity and accuracy.
• Deoxynivalenol (DON) is a prevalent mycotoxin requiring reliable detection methods.

Purpose of the Study:

• To develop a dual-mode colorimetric-SERS LFIA for sensitive and accurate deoxynivalenol (DON) detection.
• To leverage machine learning for enhanced classification and quantification of mycotoxin contamination.

Main Methods:

• Fabrication of a dual-mode LFIA using rhodium nanocores as SERS substrates and silver nanoparticles for hotspot generation.
• Optimization of electromagnetic field intensity using finite element modeling.
• Utilized Prussian blue for signal generation and background reduction.
• Applied Artificial Neural Network (ANN) and K-Nearest Neighbors (KNN) for data analysis.

Main Results:

• Achieved a detection limit of 4.21 pg/mL for DON, a 37-fold improvement over traditional colloidal gold LFIA.
• Demonstrated high classification accuracy (98.8%) and low Mean Squared Error (MSE) of 0.57 using machine learning algorithms.
• Successfully reduced background interference with Prussian blue signal at 2156 cm^-1.

Conclusions:

• The developed dual-mode LFIA platform offers superior sensitivity and accuracy for mycotoxin detection.
• Machine learning integration significantly enhances the analytical capabilities of nanosensors for complex sample analysis.
• This technology holds substantial potential for advancing harmful substance detection in various matrices.