High-throughput point-of-care serum iron testing utilizing machine learning-assisted deep eutectic solvent fluorescence detection platform

In this study, a high-throughput point-of-care testing (HT-POCT) system for detecting serum iron was developed using a hydrophobic deep eutectic solvent (HDES) fluorescence detection platform. This machine learning- assisted portable platform enables intelligent and rapid detection of trace iron ions. Blue fluorescent hydrophobic carbon quantum dots (CQDs) were synthesized using the solvothermal method. The CQDs exhibit a notable quantum yield (QY) of 36.6%, demonstrating exceptional luminescent characteristics and precise, sensitive detection capabilities for Fe3+ ions. By incorporating CQDs into specially filtered HDESs, this blend serves a dual function of concentrating iron ions from the sample and facilitating their detection. The collaboration between the two enhances the fluorescence detection signal significantly, while reducing interference from hydrophilic substances. The limit of detection can be as low as 33 nM. The principles of synthesizing HDESs and the process of extracting Fe3+ using HDESs fluorescence detection system were modeled using density functional theory (DFT). As the concentration of Fe3+ increases, the fluorescence signal detected from the sample decreases, accompanied by visible color changes when exposed to ultraviolet light. The machine learning-assisted portable platform is designed to capture fluorescence images of samples directly. The application developed using the YOLOv8 algorithm efficiently analyzes multiple samples in single or multiple images, simultaneously extracting color data from each sample and determining the concentration of iron ions. The Relative Standard Deviations (RSDs) for both single-sample and multi-sample tests were less than 10%. The machine learning-assisted portable platform provides a reliable method for detecting trace iron ions.