[19p-D102-1~6] New sensor technologies based on advanced nanobiodevices and machine learning

[19p-D102-3] Integration of Single-Molecule Analysis Methods with Machine Learning

〇Masateru Taniguchi¹, Takashi Washio¹, Tomoji Kawai¹ (1.Osaka Univ.)

Keywords:nanopores, Machine Learning

In a single-molecule analysis method based on ionic current measurements using micro/nanopore devices, a large amount of current–time graph is measured and collected. Analysis is performed using a histogram with ionic current and time as the parameters based on big data. Using this method, because the histograms of the ionic current of bacteria and viruses overlap, it is difficult to identify bacteria and viruses with high accuracy in the region where the histograms overlap. This issue has posed a challenge in developing diagnosis methods for bacteria and viruses based on ionic current measurements. Thus, the analytical method was changed to a method of analyzing individual current–time profiles via machine learning. Using this new method, a higher discrimination accuracy could be achieved with two combinations of bacteria and viruses.