Xin Li, Bing Yan
Developing a sensitive analytical platform for monitoring tiopronin (MPG), its metabolite 2-mercaptopropionic acid (MPA), and the key liver biomarker glutathione (GSH) is crucial for liver health assessment. Here, an artificial intelligence-assisted bionic vision platform based on a dual-emission Tb 3 + -functionalized hydrogen-bonded organic framework (Tb@HOF-dobpdc) is constructed. The ratiometric fluorescent sensor exhibits a distinct ″turn-on″ response, enabling highly sensitive detection of MPG, MPA, and GSH with low limits of detection (0.20, 0.68, and 0.31 μM, respectively). For practical application, Tb@HOF-dobpdc can achieve rapid detection of target analytes in real serum and urine samples by combining with the hydrogel. To overcome the human eye’s limitation in discerning subtle color changes, RGB channel processing is used to generate visually distinguishable pseudocolor signals. Furthermore, a backpropagation neural network (BPNN) is applied for accurately distinguishing analyte concentrations by identifying fluorescence images. Leveraging the strong correlation between these analytes and liver health, an AND logic gate diagnostic system is established, outputting ″healthy″ only when all biomarker levels are within safe ranges. This intelligent platform combines advanced material design and artificial intelligence, providing a powerful tool for point-of-care liver health diagnosis.