Oral cancer is a complex malignant disease of the head and neck that results from uncontrolled cell multiplication. Diabetes, alternatively, is a chronic metabolic condition characterized by elevated glucose concentration. Studies suggest that diabetes is a risk factor for oral cancer, possibly due to the overproduction of reactive oxygen species and high levels of insulin-like growth factors. Current diagnostic methods for oral cancer are time-consuming, invasive, and subject to inter-observer variability, making novel approaches necessary to identify the disease and malignancies. Fourier-transform Infrared (FTIR) spectroscopy is a sensitive and reproducible analytical technique that detects vibrational modes of molecular bonds. This study aimed to use FTIR spectroscopy to identify early biomarkers of oral cancer in diabetic patients. The study used 10 control (healthy) samples and 10 oral cancerous (diabetic) samples, with 10 independent spectra acquired from each sample to detct intra-sample heterogeneity. Spectral biomarkers that contributed to spectral variation were identified in the regions of amide I and amide III. Four machine learning models were generated, with model 1 covering the range of 600-4000cm-1, model 2 covering 1000-1800cm-1, model 3 covering 1000-1500cm-1, and
model 4 covering 1800-3000cm-1, respectively. The data was analyzed by Principal component analysis (PCA), Linear discriminant analysis (LDA), k-nearest neighbors algorithm (k-NN), and support vector machine (SVM). The highest classifier, with a prediction accuracy of over 80%, was produced by applying SVM. The results support the use of FTIR spectroscopy with machine learning as an adjunct method to the gold standard in the diagnosis of oral cancer. We continued the project with saliva samples as a case-control study, where saliva samples (80) were collected from type 1 diabetes mellitus (T1DM), type 2 diabetes mellitus (T2DM), and healthy controls (CON) were analyzed to identify specific molecular spectral signatures. Spectral analysis revealed unique vibrational modes in diabetic saliva samples compared to non-diabetic samples. Spectral biomarkers in the regions of Amide I and carbohydrates were discovered, along with subtle differences in spectra during spectral and multivariate analysis that could provide a robust and novel non-invasive
monitoring tool for diabetes. FTIR imaging has advantages over conventional
methods, as it is a label-free, non-disruptive technique requiring only a small amount of sample. It provides a biochemical fingerprinting that includes information about the structure, content, and chemical modification of any major biomolecules in the tested sample. In saliva samples, FTIR imaging provides information about modifications in proteins, DNA/RNA, and carbohydrates induced by external interventions. The spectral salivary biomarkers discovered using univariate and multivariate analysis may provide a novel and robust alternative for diabetes monitoring using non-invasive technology.