DeepRNAScanner: Deep Learning-Based Discovery of Regulatory miRNA Sequences in Lung Cancer

Lung cancer remains a formidable and life-threatening disease worldwide. Recent advancements in machine learning and artificial intelligence have led to the discovery of several significant miRNAs in lung cancer research, offering potential solutions to critical issues. In this study, we introduce miRES, a novel method that employs extreme learning machines, support vector machines, and deep learning techniques to identify lung cancer-associated miRNAs. Our approach tackles the inherent two-class classification problem using machine learning and deep learning algorithms. By leveraging the miRES method, we uncover detailed information about miRNA sequences and discern their regulatory functions, including up- and down-regulation. Through comparative analysis with other machine learning and deep learning methods, we demonstrate that miRES surpasses performance metrics such as sensitivity, specificity, and accuracy. Specifically, miRES achieve an impressive 83.34% sensitivity (Sn), 78.55% specificity (Sp), 0.73 F1 score, and 0.4882 Matthews correlation coefficient (MCC). Furthermore, our proposed method outperforms various deep learning methods, including CNN, ResNet101, ResNet152, VGG16, VGG19, AlexNet, and GoogleNet, in classifying miRNA sequences, attaining a K-10 value of 96.25%. Overall, miRES offers a powerful and effective approach for identifying and characterizing lung cancer miRNA sequences, leveraging the strengths of machine learning and deep learning techniques. This research contributes to the growing body of knowledge aimed at combating lung cancer and improving patient outcomes.