Early detection of head and neck cancer (HNC) is critical for improving prognosis and survival rates. Many cases are diagnosed at advanced stages due to subtle initial symptoms and the complexity of the head and neck anatomy, which complicates clinical examination and biopsy procedures. Therefore, there is an urgent need for non-invasive, reliable, and rapid diagnostic methods suitable for point-of-care (POC) settings. In this study, we applied surface-enhanced Raman spectroscopy (SERS) to develop a rapid screening method for HNC diagnosis using cerumen as the medium. The study aimed to utilize a SERS-based machine learning (ML) approach to distinguish between cerumen samples from healthy individuals and those with HNC. Principal Component Analysis and Linear Discriminant Analysis (PCA-LDA) were performed to analyze and differentiate the cerumen samples. By comparing the SERS spectra of healthy donors with those of HNC patients, we identified SERS spectral features associated with the presence of tumors. The PCA-LDA method successfully classified healthy and HNC cerumen samples with 87.2% accuracy, 87.3% specificity, 87% sensitivity, and a 90% area under the receiver operating characteristic (ROC AUC) curve. This cerumen-SERS-ML workflow proved effective for the rapid identification and evaluation of HNC, offering a promising tool for disease diagnosis.