Machine learning-empowered nanobody-based dual-mode immunosensor using gold-silver nanoclusters for the accurate detection of Shigella flexneri in food.

Sensitive and rapid detection of Shigella flexneri (S. flexneri) is of great significance for food safety and public health. In this study, a nanobody-based colorimetric/fluorescent dual-mode immunosensor was developed for the first time, and a machine learning model was integrated for intelligent fusion of the dual signals. A phage-displayed nanobody library was constructed by immunizing Bactrian camels with formalin-inactivated S. flexneri and specific nanobodies were obtained after four rounds of solid-phase panning. Using bovine serum albumin-protected gold-silver bimetallic nanoclusters (Au-Ag NCs) as fluorescent probes, a dual-signal output platform was established. Under optimized conditions, the colorimetric mode achieved a limit of detection (LOD) of 2.992 × 104 CFU/mL, and exhibited no cross-reactivity with other common foodborne pathogens. The LOD in the fluorescent mode was as low as 4.074 × 103 CFU/mL, which is 7.3 times lower than that of the colorimetric mode. A machine learning model combining K-Nearest Neighbors classification and Random Forest regression was employed for intelligent calibration of the dual signals, significantly improving the reliability and stability of the assay. With the assistance of the machine learning model, the coefficient of determination (R2) between predicted and actual concentrations reached 0.9953. Spike-and-recovery experiments in food samples yielded recovery rates ranging from 81.53% to 109.46%, demonstrating high accuracy. This study provides a novel, rapid, accurate, and environmentally friendly strategy for the detection of S. flexneri, holding great promise for applications in food safety monitoring.