Abstract
Gastric cancer represents a significant public health challenge, necessitating advancements in early diagnostic methodologies. This investigation employed attenuated total reflection Fourier-transform infrared (ATR-FTIR) spectroscopy to conduct a multivariate analysis of human serum. The study encompassed the examination of blood samples from 96 individuals diagnosed with gastric cancer and 96 healthy volunteers. Principal component analysis (PCA) was utilized to interpret the infrared spectral data of the serum samples. Specific spectral bands exhibiting intensity variations between the two groups were identified. The infrared spectral ranges of 3500 ~ 3000 cm⁻(1), 1700 ~ 1600 cm⁻(1), and 1090 ~ 1070 cm⁻(1) demonstrated significant diagnostic value for gastric cancer, likely attributable to differences in protein conformation and nucleic acids. By employing machine learning algorithms to differentiate between gastric cancer patients (n = 96) and healthy controls (n = 96), we achieved a sensitivity of up to 89.7% and a specificity of 87.2%. Receiver operating characteristic (ROC) analysis yielded an area under the curve (AUC) of 0.901. These findings underscore the potential of our serum-based ATR-FTIR spectroscopy examination method as a straightforward, minimally invasive, and reliable diagnostic test for the detection of gastric cancer.