Artificial intelligence-enabled serum SERS biomolecular fingerprinting for non-invasive stratification of gastric...

Gastric cancer (GC) is one of the leading causes of cancer-related mortality worldwide due to its late-stage diagnosis and progressive metastatic behavior. In this research, surface-enhanced Raman spectroscopy (SERS) coupled with artificial intelligence-based chemometric analysis was employed as a rapid, non-invasive platform for the stage-wise discrimination of GC using blood serum samples. Silver nanoparticles (AgNPs) were utilized as a SERS-active substrate to enhance the Raman signals of biomolecular constituents in serum samples. The mean SERS spectra recorded in the spectral region of 400-1600 cm^-1 revealed significant biochemical changes associated with proteins, nucleic acids, lipids, and amino acids during GC progression from Stage 1 to Stage 4. Characteristic Raman bands observed near 493, 723, 884, 1002, 1130, and 1390 cm^-1 indicated progressive metabolic and structural variations associated with disease progression. Principal component analysis (PCA) and hierarchical cluster analysis (HCA) demonstrated clear biochemical discrimination between healthy controls and different pathological stages of GC. Furthermore, PCA-assisted support vector machine (PCA-SVM) and PCA-assisted k-nearest neighbor (PCA-KNN) models were developed to compare classification performance. The PCA-KNN model demonstrated excellent stage-wise discrimination with an optimal k value of 5 and a maximum cross-validation accuracy of 94.03%. Receiver operating characteristic (ROC) analysis revealed high classification efficiency with area under the curve (AUC) values of 0.973, 0.921, 0.986, 0.967, and 0.956 for Healthy, Stage 1, Stage 2, Stage 3, and Stage 4 groups, respectively. The PCA-SVM model achieved comparatively robust classification performance for differentiating closely related pathological stages. These findings demonstrate that serum SERS integrated with machine learning algorithms provides a rapid, sensitive, and non-invasive analytical tool for GC detection and progression monitoring.