Integrating miRNA profiling and machine learning for improved prostate cancer diagnosis.

Prostate cancer (PCa) diagnosis remains challenging due to overlapping clinical features with benign prostatic hyperplasia (BPH) and limitations of existing diagnostic tools like PSA tests, which yield high false-positive rates. This study investigates the potential of microRNA (miRNA) biomarkers, analyzed via reverse transcription polymerase chain reaction and machine learning (ML), to enhance diagnostic accuracy. miRNAs such as miR-21-5p, miR-141-3p, and miR-221-3p were identified as significant discriminators between PCa and BPH through a prospective cohort study. Whole blood miRNA profiling offered a robust systemic representation of disease states. A random forest ML model was trained on expression data, achieving notable performance metrics: an accuracy of 77.42%, AUC of 0.78 during verification, and 74.07% accuracy and 0.75 AUC in validation. The model's use of miRNA expression ratios, such as miR-141-3p/miR-221-3p, demonstrated superior sensitivity and specificity over traditional PSA testing. Bioinformatics analysis confirmed the association of selected miRNAs with cancer pathways, including PD-L1/PD-1 checkpoint and androgen receptor signaling, validating the biological relevance of the findings. This novel integration of miRNA profiling and machine learning holds great potential for the clinical translation of miRNA-based non-invasive diagnostics, enhancing diagnostic precision. However, broader population studies and standardization of protocols are needed to ensure scalability and clinical applicability. This research provides a foundational framework for advancing miRNA-based diagnostics, bridging discovery and clinical implementation.