Integrating Bioinformatics and Machine Learning for Gastric Cancer Biomarker Discovery and Diagnostic Modelling

Abstract

Background: Gastric cancer remains one of the leading causes of cancer-related mortality worldwide, highlighting the urgent need for novel biomarkers and therapeutic targets. Objectives: This study integrates bioinformatics and machine learning to analyze RNA sequencing data from the Gene Expression Omnibus (GEO) dataset GSE164174, focusing on dysregulated microRNAs (miRNAs) and their roles in gastric carcinogenesis. Methods: Differential expression analysis identified 2,565 significantly altered miRNAs, including hsa-miR-4481 [upregulated, log2 fold change (logFC) = 2.23] and hsa-miR-5100 [downregulated, logFC = -3.91]. Pathway enrichment revealed miRNA-mediated mRNA destabilization (GO:0035279, P = 4.20E-63) as a critical mechanism. Machine learning models, including random forests, were used to assess diagnostic accuracy for top miRNAs. Results: Machine learning models demonstrated high diagnostic accuracy [area under the curve (AUC) > 0.90] for top miRNAs. These findings underscore the potential of miRNAs as non-invasive biomarkers and therapeutic targets. Conclusions: By bridging computational predictions with biological insights, this study provides a foundation for clinical translation and personalized oncology.