Decision Tree and Linear Discriminant Analysis-Assisted Design of Polydopamine Nanoparticle-Based Colorimetric Array Sensor Modulated by Metal Ions for High-Efficiency Detection of Physiological Phosphates

Physiological phosphates play a crucial role in various mechanisms of biological processes and diseases. Their interconversion through enzymatic reactions reveals the importance of developing efficient methods for evaluating physiological phosphates and their complex mixtures. Herein, we prepared three polydopamine nanoparticles (PDA NPs) with various surface modifications via Michael addition reactions between dopamine and different biothiols. A high-dimensional dual-signal array sensor comprising three PDA NPs, nine metal ions, and 3,3',5,5'-tetramethylbenzidine (TMB) was preliminarily designed. Decision tree and linear discrimination analysis (LDA) algorithms were employed as core design tools to identify the most effective sensing units, enabling the development of a streamlined N-acetyl-l-cysteine PDA NPs (N-PDA NPs)-Fe³⁺, Ag⁺, Hg²⁺-TMB colorimetric array sensor for phosphates analysis. Competitive binding between phosphates and N-PDA NPs with metal ions induced differential changes in the amount of TMB oxidation product (oxTMB), generating unique fingerprint response patterns. This sensor successfully discriminated seven phosphates (ATP, ADP, AMP, PPi, HPO₄^2-, H₂PO₄^-, and Pi) at equal concentrations as well as common phosphates at varying concentrations and their mixtures. A broader range of phosphate species quantification and unknown phosphates accurate prediction validated sensor's robustness. Notably, the sensor realized the monitoring of kinase activity involving phosphates. The 100% discrimination of phosphates from common interfering substances and complete differentiation of them in complex matrix demonstrated sensor's high selectivity. The recovery rate for phosphate in spiked serum ranged from 96% to 114%. More importantly, this study represented the first successful establishment of an intuitive evaluation model for serum phosphate levels, highlighting potential for diagnosing phosphate-related diseases.