Lung Cancer Stages Classification Based on Differential Gene Expression and Deep Learning

Lung cancer, a malignancy originating from the epithelial cells lining the lung airways, ranks as a leading cause of cancer-related mortality across genders. Among the various histological subtypes of non-small cell lung cancer (NSCLC), lung adenocarcinoma (LUAD) and squamous cell carcinoma (LUSC) are prominent. Discriminating between healthy and diseased states at the molecular level necessitates the identification of differentially expressed genes (DEGs). Biomedical literature and microarray investigations serve as vital resources for elucidating gene expression variations. A gene is considered highly expressed when a notable statistical difference or alteration in read counts or expression level indices is observed between distinct experimental conditions. A sound gene expression strategy lays the groundwork for effective gene silencing interventions, thereby impeding or halting cancer progression. This study undertakes a rigorous analysis of messenger ribonucleic acid (mRNA) data derived from LUAD samples to identify the most differentially expressed genes. The first model applies deep learning methodologies to perform feature extraction on the data, followed by classification into the various stages of lung cancer, achieving an impressive accuracy of 99%. The second model embarks on an exploratory analysis using the DESeq2 package to identify differentially expressed genes, subsequently conducting an enrichment analysis to elucidate the biological and molecular pathways implicated in lung cancer. Comparative evaluations against state-of-the-art methods demonstrate the models’ high accuracy and efficacy in discerning intricate molecular signatures and unveiling potential lung cancer biomarkers.