Machine learning-assisted dual-mode hydrogel sensor for the detection of nitrite

Abstract

Accurate on-site detection of nitrite (NO₂⁻) is essential for ensuring food safety and human health. In this study, we developed a portable dual-mode hydrogel sensor combining ratiometric fluorescence and colorimetric detection for accurate nitrite monitoring in food samples. The sensor was fabricated by embedding graphene quantum dots (GQDs) into an agarose hydrogel matrix. Blue-emissive GQDs (λ = 450 nm) functioned as a ratiometric reference and were effectively quenched by orange-emissive benzotriazole (BTA, λ = 578 nm), which forms via a diazo coupling reaction between nitrite and o-phenylenediamine (OPD). This reaction enabled a nitrite-dependent ratiometric fluorescence response, accompanied by a yellow color intensification resulting from the formed BTA. For field applications, a portable hydrogel kit was constructed and integrated with a back-propagation artificial neural network (BPANN) to facilitate the interpretation of nitrite-induced signal changes. The sensor demonstrated high sensitivity, with detection limits of 2.1 × 10⁻³ mg/L in fluorescence mode, 0.088 mg/L in colorimetric mode, and 0.07 mg/L in the hydrogel-based system. This portable platform enables rapid (<15 min), AI-assisted analysis with cross-platform compatibility, offering a promising tool for food and environmental safety monitoring.