Machine Learning-Enhanced Analysis of miRNA Biomarkers for Accurate Breast Cancer Diagnosis Using DNA Seagrass

As potential biomarkers for breast cancer, microRNAs (miRNAs) have demonstrated significant promise in clinical applications. However, accurate miRNA-based breast cancer diagnosis is hindered by the lack of simple, ultrasensitive, and highly specific detection methods and reliable biomarkers. To tackle these challenges, we introduced an innovative strategy using rolling circle amplification-generated DNA seaweed (RCA-GDS) to detect the multiple miRNA biomarkers combined with machine learning to enable precise breast cancer diagnosis. RCA-GDS effectively converts linear RCA amplification into exponential amplification, efficiently enhancing fluorescence signals and enabling the detection of miRNAs at concentrations as low as attomolar levels within 2 h under isothermal conditions. Using the TCGA database, we screened a panel of miRNAs (miRNA21, miRNA182, and miRNA183) for the precise diagnosis of breast cancer and validated their reliability in both intracellular and serum samples. Finally, we integrated machine learning algorithms with the miRNA detection system to develop a differential diagnosis model, which was further validated in an independent cohort and demonstrated excellent diagnostic accuracy. This work not only enables ultrasensitive and highly specific miRNA detection but also advances miRNA panel-based clinical applications in breast cancer diagnosis.