Machine Learning-Aided Intelligent Monitoring of Multivariate miRNA Biomarkers Using Bipolar Self-powered Sensors

Abstract

Breast cancer has become the most prevalent form of cancer among women on a global scale. The early and timely diagnosis of breast cancer is of the utmost importance for improving the survival rate of patients with this disease. The occurrence of breast cancer is typically accompanied by the dysregulation of multiple microRNA (miRNA) expression profiles. Consequently, simultaneous detection of multiple miRNAs is vital for the early and accurate diagnosis of breast cancer. In this study, a bipolar self-powered sensor was developed for the simultaneous detection of miRNA-451 and miRNA-145 breast cancer biomarkers based on the specific catalytic properties of enzymes. Selenides with a microporous hollow cubic structure were designed and prepared, which can markedly enhance the enzyme load and activity, as well as detection sensitivity, due to their extensive surface area and three-dimensional porous channel. The designed bipolar self-powered sensor platform is integrated into the commercial chip, and the signal is presented in the smartphone interface, thereby enabling real-time and continuous monitoring. Furthermore, machine learning was utilized to predict miRNA detection, which encompasses numerous stages, including data collection, feature extraction, model training, and validation. In comparison to the limited sensing efficiency of self-powered biosensors driven by enzyme biofuel cells, our bipolar self-powered sensor achieved simultaneous quantitative analysis of multiple miRNA targets, thereby providing a robust tool for a more comprehensive understanding of miRNA function and its association with cancers.