A novel machine learning-based predictive model for gastric cancer.

[BACKGROUND] Gastric cancer (GC) is a prevalent malignancy worldwide, necessitating the discovery of biomarkers for early diagnosis and progression prediction. This study aimed to identify core genes associated with GC.

[METHODS] This study integrated data from the Gene Expression Omnibus (GEO) database, encompassing multiple datasets. Differential expression and enrichment analyses identified genes linked to GC. Using machine learning algorithms-least absolute shrinkage and selection operator (LASSO) regression, support vector machine (SVM), and random forest (RF)-predictive models were constructed, with the optimal one selected for further investigation. The SHapley Additive exPlanations (SHAP) method was applied to assess the contribution of core genes. Additionally, gene set enrichment analysis (GSEA) and immune cell infiltration analysis were conducted to explore related molecular mechanisms.

[RESULTS] This study identified 130 differentially expressed genes (DEGs), which exhibited enrichment in functions and pathways potentially linked to GC. Through the collective application of multiple machine learning methods, 4 key genes associated with GC (, , , and ) were pinpointed. The RF model, demonstrating superior accuracy, was chosen for subsequent SHAP analysis to elucidate the contributions of these genes. Furthermore, GSEA and immune cell infiltration analysis revealed distinct molecular and immune profiles between GC and normal tissues.

[CONCLUSIONS] This study provided potential biomarkers and contributed to the theoretical basis for GC prevention and treatment.