Machine Learning-Assisted Ratiometric Fluorescence Electrospun Nanofiber Films for Portable and Intelligent...

The intelligent authentication of whole wheat products remains a significant challenge due to the difficulty in simultaneously monitoring multiple alkylresorcinol (AR) homologues within complex food matrices. To address this, we have developed a novel sensing platform integrating machine learning (ML) algorithms with advanced ratiometric fluorescence (FL) materials. The core component is a dual-emission g-C₃N₄/Ru, leveraging the blue fluorescence from g-C₃N₄ nanosheets as the analytical signal and the red fluorescence from [Ru(bpy)₃]²⁺ as an internal reference. Upon interaction with AR homologues, a visible color change from blue-violet to pink occured due to multiple synergistic effects of IFE, a-PET, electrostatic attraction, and π-π interactions. The system exhibited exceptional sensitivity for quantitative detection of five AR homologues (C17:0, C19:0, C21:0, C23:0, C25:0) within the concentration range of 0-60 μg·mL^-1, achieving ultralow limit of detections (LODs) ranging from 2.1 to 9.9 ng·mL^-1. For precise and portable quantitative analysis, a random forest-back-propagation neural network (RF-BPNN) algorithm-assisted FL electrospun film (RuCN PAN NFs) was employed, enabling reliable on-site monitoring. RGB features were extracted using the OpenCV library, and 252 samples were evenly divided through stratified sampling to maintain data balance, yielding exceptional prediction accuracy (R² = 0.9822) and robust performance by training with the RF-BPNN. Application to commercial wheat samples confirms the system's utility for AR detection and whole wheat authentication. This integrated approach overcomes traditional analytical limitations by combining ML algorithms with advanced materials, enabling intelligent, rapid, and on-site detection of AR homologues. It provides a promising tool for verifying the authenticity of whole wheat products, offering significant potential for food safety monitoring and quality control applications.