Artificial Intelligence-Assisted Multiplex Detection of Organophosphate Flame Retardants via Cataluminescence Principle

The precise multiplex detection of chemical species is of significant importance in the field of analytical chemistry. However, the current methods for multiplex detection have been rather limited. Herein, an artificial intelligence-assisted cataluminescence (AI-assisted CTL) multiplex detection strategy is creatively developed. Organophosphate flame retardants (OPFRs), a typical class of concerned emerging contaminants, were chosen as the targets. The CTL dynamic curves of OPFRs, which encompassed abundant thermodynamic and kinetic information, were collected as the data set and further preprocessed for noise reduction via filtering treatment. Subsequently, a back-propagation (BP) artificial neural network combined with principal component analysis (PCA) was constructed to extract the characteristic parameters and successfully realize the multiplex recognition. The results showed that this optimized PCA-BP model enabled the distinction of six representative OPFRs and their mixtures as well as evaluating the constituent concentrations, which achieved an average classification accuracy of 98.38% and quantitative accuracy of 94.42%. Briefly, this AI-assisted CTL method demonstrated an excellent ability to simultaneously determine the number, types, and concentrations of target analytes in an unknown media environment. Notably, the superiority of this established PCA-BP model was further confirmed through the comparative analysis with other learning algorithms. This proposed innovative analytical strategy represents a profound integration of CTL sensing technology with AI and put provides effective guidance for the interdisciplinary fusion between chemical sensing and information technology.