Deep Learning-Assisted Dual-Mode Lanthanide MOF/Anthocyanins Platform for Real-Time Pork Freshness Sensing

Real-time and accurate monitoring of meat freshness remains a critical challenge for intelligent food-sensing systems. Herein, a dual-mode sensor array was developed by combining ratiometric fluorescent lanthanide metal-organic frameworks (Ln-MOFs) with pH-responsive anthocyanins for real-time, high-precision monitoring of pork freshness. Specifically, two "traffic-light" ratiometric fluorescent probes were engineered by doping Eu³⁺ and Tb³⁺ into UiO-66-OH. These probes exhibited exceptional sensitivity and specificity toward ammonia and biogenic amines, achieving detection limits for ammonia as low as 1.224 ppm (Eu³⁺-doped) and 2.359 ppm (Tb³⁺-doped). Furthermore, the dual-mode array demonstrated robust fluorescence-colorimetric responses and outstanding color stability (Euclidean distance <5 after 14 days of storage). During pork freshness monitoring, fluorescence signals enabled sensitive early-spoilage warnings at TVB-N levels of 15-20 mg/100 g (less fresh state), while colorimetric signals provided visual cues for naked-eye discrimination, changing from pink to purple and then to blue. By coupling the array with three convolutional neural networks, high-accuracy classification of pork into fresh, less fresh, and spoiled states was achieved, reaching a maximum accuracy of 97.8%. This strategy provides a reliable and early-warning-capable solution for real-time pork freshness monitoring in intelligent food packaging and logistics.