Machine Learning‐Coupled Upconversion Nanoprobe‐Based Immunochromatographic Device for Blood Drug Dynamics Monitoring

Abstract The dynamic concentration of drugs in blood reflects pharmacokinetics and is a key indicator of therapeutic efficacy and safety, emphasizing the need for real‐time monitoring to guide individualized dosing and reduce adverse effects. Herein, a ratiometric fluorescent sensor is developed by functionalizing upconversion nanoparticles (UCNPs) with p‐dimethylaminobenzaldehyde (p‐DMAB) and integrating the resulting probe into a portable immunochromatographic device for the sensitive and rapid analysis of isoniazid dosage in whole blood. The unique properties of UCNPs, including near‐infrared excitation, low background interference, and high photostability, together with the probe's high chemical specificity, enabling direct detection in whole blood without complex pretreatment. Upon a selective chemical reaction between isoniazid and the p‐DMAB‐modified UCNPs, the sensor exhibits a concentration‐dependent fluorescence color change from blue to red, allowing clear discrimination of therapeutically relevant dosage ranges. To eliminate subjective visual errors and enable batch analysis, the platform integrates 3D printing with smartphone‐based imaging and convolutional neural network modeling. The CNN algorithm quantitatively maps fluorescence intensity ratios to drug concentrations, effectively removing operator bias. This intelligent, low‐cost, and user‐friendly system offers a reliable solution for point‐of‐care therapeutic drug monitoring and can be adapted for other clinical drug targets, supporting personalized medicine and decentralized healthcare.