Machine Learning-Assisted Portable Ai BOX Based on RPA-CRISPR/Cas12a for Rapid On-Site Detection of Foodborne...

Foodborne pathogens present a major threat to global public health. However, conventional detection methods and equipment are often unsuitable for the on-site and timely monitoring of these pathogens. To overcome this critical limitation and establish a rapid detection workflow, we developed the portable smart Ai BOX (artificial intelligence BOX). This device is a compact, palm-sized, internet of things (IoT)-enabled instrument that utilizes isothermal fluorescence diagnostics and weighs only 180 g. The Ai BOX features an optimized minimalist industrial design, ultralow power consumption, and a high-sensitivity optical sensing system. The device performs real-time fluorescence detection, with results automatically interpreted and transmitted to a dedicated mobile application (APP) via an integrated smart camera, enabling comprehensive food monitoring. Furthermore, the incorporation of artificial intelligence and machine learning (ML) algorithms significantly enhances the processing capability of the RPA-CRISPR/Cas12a fluorescence signal, thereby ensuring superior detection accuracy. The Ai BOX is ideally suited for on-site point-of-care testing (POCT) of foodborne pathogens. By integrating the one-pot-RPA-CRISPR/Cas12a method, the device achieves an exceptionally low limit of detection (LOD) of 1 × 10¹ CFU/mL for Listeria monocytogenes. In tests using simulated samples, it demonstrated 100% sensitivity and specificity. Consequently, the Ai BOX exhibits promising application potential for diverse public and personal health scenarios, including the detection of meat adulteration, food contamination, and wastewater monitoring.