Different data analysis models for detecting miRNAs using nanopores

Nanopore sensors generate complex ionic current traces that encode rich molecular information, making them well-suited for computational signal analysis. Building on previous work in which DNA-barcoded probes were developed for the multiplexed detection of biomarkers, this study explores different analytical approaches for interpreting binary nanopore signals. The method relies on identifying two classes of events: (1) non-delayed signals, indicating the absence of a target biomarker, and (2) delayed signals, resulting from probe-target interactions, proving the presence of the biomarker. To classify these events and enable robust biomarker detection, we applied three different computational strategies: (1) moving standard deviation (MSD), (2) spectral entropy (SE), and (3) a convolutional neural network (CNN) model. While MSD and SE represent conventional, rule-based approaches requiring manually defined thresholds, the CNN model offers a more scalable, automated solution for classifying nanopore events directly from raw time series data. All three methods were capable of distinguishing between signal classes; however, the CNN model demonstrated superior sensitivity and robustness. This work highlights the importance of data interpretation in nanopore sensing and presents a comparative framework for binary event classification. The findings pave the way for the development of machine learning-driven nanopore diagnostics capable of detecting diverse biomarker types at the single-molecule level.