Development of Phenotypic Microbial Sensing System by AI-driven AC Nanopore Method

In this study, we developed a double-sided offset ring-electrode nanopore device and a prototype hand-held measurement system, and evaluated the overall performance for the application of AI-driven AC nanopore method to microbial phenotypic sensing. First, the particle measurement performance was evaluated and found that the sensitivity was approximately three times higher than that of conventional systems for particles up to 120 nm in diameter. n microbial identification using a single nanopore device, the accuracy of classification of 16 types of bacteria and 11 types of viruses using a convolutional neural network was approximately 83% and 78%, respectively. This suggests the possibility of classifying both bacteria and viruses with a single nanopore diameter. Furthermore, the method achieved 94.5% accuracy in three basic classification tasks of conventional colony counting method, demonstrating that this method is upward compatible with colony counting methods.