Automated-Screening Oriented Electric Sensing of Vitamin B1 Using a Machine Learning Aided Solid-State Nanopore

Micronutrient detection and identification at the single-molecule level are paramount for both clinical and home diagnostics. Analytical tools such as high-performance liquid chromatography and liquid chromatography-tandem mass spectrometry have been widely used but include a high instrument cost and prolonged analysis time. Here, as a model system, by merging nanopore signatures with machine learning algorithms, we propose an automated electric sensing strategy to identify vitamin B1 and its phosphorylated derivatives with good accuracy. Further, the relationship between vitamin B1 dynamics and nanopore signatures is examined. To understand the machine-decision-making process, Shapley additive explanations are made. Using a machine learning aided solid-state nanopore, we pave the way for next-generation micronutrient detection.